Spaces:
Runtime error
Runtime error
Delete MiniGPT-Med-github
Browse filesThis view is limited to 50 files because it contains too many changes. Β
See raw diff
- MiniGPT-Med-github/Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.1495.1517874291.249176.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.16254.1517874395.786150.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/1.2.840.113654.2.55.48339325922382839066544590341580673064.png +0 -0
- MiniGPT-Med-github/Med_examples_v2/1.3.6.1.4.1.14519.5.2.1.7009.9004.242286124999058976921785904029.png +0 -0
- MiniGPT-Med-github/Med_examples_v2/5f4e8079-8225a5d2-1b0c3c46-4394a094-f285db0e.jpg +0 -3
- MiniGPT-Med-github/Med_examples_v2/synpic33889.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/synpic50958.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/synpic56061.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/synpic58547.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/synpic60423.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/synpic676.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/xmlab149/source.jpg +0 -0
- MiniGPT-Med-github/Med_examples_v2/xmlab589/source.jpg +0 -0
- MiniGPT-Med-github/README.md +0 -51
- MiniGPT-Med-github/dcgm/bash/34649895/dcgm-gpu-stats-gpu202-02-r-34649895.out +0 -39
- MiniGPT-Med-github/dcgm/bash/34673507/dcgm-gpu-stats-gpu201-23-l-34673507.out +0 -39
- MiniGPT-Med-github/dcgm/bash/34676162/dcgm-gpu-stats-gpu201-23-l-34676162.out +0 -39
- MiniGPT-Med-github/dcgm/bash/34691276/dcgm-gpu-stats-gpu201-09-l-34691276.out +0 -42
- MiniGPT-Med-github/dcgm/bash/34709014/dcgm-gpu-stats-gpu109-16-l-34709014.out +0 -39
- MiniGPT-Med-github/dcgm/bash/34721198/dcgm-gpu-stats-gpu203-23-r-34721198.out +0 -57
- MiniGPT-Med-github/dcgm/bash/34734121/dcgm-gpu-stats-gpu201-23-l-34734121.out +0 -35
- MiniGPT-Med-github/dcgm/bash/34738689/dcgm-gpu-stats-gpu201-16-r-34738689.out +0 -35
- MiniGPT-Med-github/dcgm/bash/34757693/dcgm-gpu-stats-gpu202-16-r-34757693.out +0 -42
- MiniGPT-Med-github/demo_v2.py +0 -648
- MiniGPT-Med-github/environment.yml +0 -35
- MiniGPT-Med-github/eval_configs/minigptv2_benchmark_evaluation.yaml +0 -69
- MiniGPT-Med-github/eval_configs/minigptv2_eval.yaml +0 -24
- MiniGPT-Med-github/eval_scripts/.DS_Store +0 -0
- MiniGPT-Med-github/eval_scripts/__pycache__/IoU.cpython-39.pyc +0 -0
- MiniGPT-Med-github/eval_scripts/__pycache__/clean_json.cpython-39.pyc +0 -0
- MiniGPT-Med-github/eval_scripts/__pycache__/metrics.cpython-39.pyc +0 -0
- MiniGPT-Med-github/eval_scripts/clean_json.py +0 -74
- MiniGPT-Med-github/eval_scripts/metrics.py +0 -164
- MiniGPT-Med-github/eval_scripts/model_evaluation.py +0 -274
- MiniGPT-Med-github/miniGPTV2.yml +0 -35
- MiniGPT-Med-github/miniGPT_Med_.pth +0 -3
- MiniGPT-Med-github/minigpt4/.DS_Store +0 -0
- MiniGPT-Med-github/minigpt4/__init__.py +0 -31
- MiniGPT-Med-github/minigpt4/__pycache__/__init__.cpython-310.pyc +0 -0
- MiniGPT-Med-github/minigpt4/__pycache__/__init__.cpython-39.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/.DS_Store +0 -0
- MiniGPT-Med-github/minigpt4/common/__init__.py +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/__init__.cpython-310.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/__init__.cpython-39.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/config.cpython-310.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/config.cpython-39.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/dist_utils.cpython-310.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/dist_utils.cpython-39.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/eval_utils.cpython-39.pyc +0 -0
- MiniGPT-Med-github/minigpt4/common/__pycache__/logger.cpython-310.pyc +0 -0
MiniGPT-Med-github/Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.1495.1517874291.249176.jpg
DELETED
Binary file (63.8 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.16254.1517874395.786150.jpg
DELETED
Binary file (54.6 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/1.2.840.113654.2.55.48339325922382839066544590341580673064.png
DELETED
Binary file (162 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/1.3.6.1.4.1.14519.5.2.1.7009.9004.242286124999058976921785904029.png
DELETED
Binary file (142 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/5f4e8079-8225a5d2-1b0c3c46-4394a094-f285db0e.jpg
DELETED
Git LFS Details
|
MiniGPT-Med-github/Med_examples_v2/synpic33889.jpg
DELETED
Binary file (22.3 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/synpic50958.jpg
DELETED
Binary file (35.9 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/synpic56061.jpg
DELETED
Binary file (40.7 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/synpic58547.jpg
DELETED
Binary file (47.2 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/synpic60423.jpg
DELETED
Binary file (46.7 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/synpic676.jpg
DELETED
Binary file (55 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/xmlab149/source.jpg
DELETED
Binary file (20.9 kB)
|
|
MiniGPT-Med-github/Med_examples_v2/xmlab589/source.jpg
DELETED
Binary file (22 kB)
|
|
MiniGPT-Med-github/README.md
DELETED
@@ -1,51 +0,0 @@
|
|
1 |
-
# MiniGPT-Med: Large Language Model as a General Interface for Radiology Diagnosis
|
2 |
-
Asma Alkhaldi, Raneem Alnajim, Layan Alabdullatef, Rawan Alyahya, Jun Chen, Deyao Zhu, Ahmed Alsinan, Mohamed Elhoseiny
|
3 |
-
|
4 |
-
*Saudi Data and Artificial Intelligence Authority (SDAIA) and King Abdullah University of Science and Technology (KAUST)*
|
5 |
-
|
6 |
-
## Installation
|
7 |
-
```
|
8 |
-
git clone https://github.com/Vision-CAIR/MiniGPT-Med
|
9 |
-
cd MiniGPT-Med
|
10 |
-
conda env create -f environment.yml
|
11 |
-
conda activate miniGPT-Med
|
12 |
-
```
|
13 |
-
|
14 |
-
## Download miniGPT-Med trained model weights
|
15 |
-
|
16 |
-
* miniGPT-Med's weights [miniGPT-Med Model](https://drive.google.com/file/d/1kjGLk6s9LsBmXfLWQFCdlwF3aul08Cl8/view?usp=sharing)
|
17 |
-
|
18 |
-
* Then modify line 8 at miniGPT-Med/eval_configs/minigptv2_eval.yaml to be the path of miniGPT-Med weight.
|
19 |
-
|
20 |
-
## Prepare weight for LLMs
|
21 |
-
|
22 |
-
### Llama2 Version
|
23 |
-
|
24 |
-
```shell
|
25 |
-
git clone https://huggingface.co/meta-llama/Llama-2-13b-chat-hf
|
26 |
-
```
|
27 |
-
|
28 |
-
Then modify line 14 at miniGPT-Med/minigpt4/configs/models/minigpt_v2.yaml to be the path of Llama-2-13b-chat-hf.
|
29 |
-
|
30 |
-
## Launching Demo Locally
|
31 |
-
|
32 |
-
```
|
33 |
-
python demo.py --cfg-path eval_configs/minigptv2_eval.yaml --gpu-id 0
|
34 |
-
```
|
35 |
-
|
36 |
-
## Dataset
|
37 |
-
| Dataset | Images | json file|
|
38 |
-
|---------|---------|----------|
|
39 |
-
| MIMIC |[Download](https://physionet.org/content/mimiciii/1.4/) | [Download](https://drive.google.com/drive/folders/1nZhdfNoh7fkx7CWvf0_47_OLv3tA2m3o?usp=sharing) |
|
40 |
-
| NLST |[Download](https://wiki.cancerimagingarchive.net/display/NLST)| [Downlaod](https://drive.google.com/drive/folders/1OKgMTaGLu_dWRuco6JipYzezw3oNwgaz?usp=sharing) |
|
41 |
-
|SLAKE |[Downlaod](https://www.med-vqa.com/slake/) |[Download](https://drive.google.com/drive/folders/1vstjmfRbKahSAsi_b6FmTQiuolvgO8oC?usp=sharing)|
|
42 |
-
|RSNA |[Downlaod](https://www.rsna.org/rsnai/ai-image-challenge/rsna-pneumonia-detection-challenge-2018) | [Download](https://drive.google.com/drive/folders/1wkXPvUNqda6jWAIduyiVJkS3Tx7P7td8?usp=sharing) |
|
43 |
-
|Rad-VQA |[Downalod](https://osf.io/89kps/) |[Download](https://drive.google.com/drive/folders/1ING6Dodwk2DU_t4GHQYudNFMMg9OMfBQ?usp=sharing) |
|
44 |
-
|
45 |
-
## Acknowledgement
|
46 |
-
|
47 |
-
- MiniGPT-4
|
48 |
-
- Lavis
|
49 |
-
- Vicuna
|
50 |
-
- Falcon
|
51 |
-
- Llama 2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34649895/dcgm-gpu-stats-gpu202-02-r-34649895.out
DELETED
@@ -1,39 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34649895.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 0 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Tue Jul 9 09:29:46 2024 |
|
7 |
-
| End Time | Wed Jul 10 09:30:32 2024 |
|
8 |
-
| Total Execution Time (sec) | 86445.3 |
|
9 |
-
| No. of Processes | 1 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 232291 |
|
12 |
-
| Power Usage (Watts) | Avg: 65.6704, Max: 84.315, Min: 61.555 |
|
13 |
-
| Max GPU Memory Used (bytes) | 10104078336 |
|
14 |
-
| SM Clock (MHz) | Avg: 595, Max: 1155, Min: 210 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 1548651 |
|
34 |
-
| Avg SM Utilization (%) | 0 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
+----- Overall Health -------------+-----------------------------------------+
|
37 |
-
| Overall Health | Healthy |
|
38 |
-
+------------------------------------+-----------------------------------------+
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34673507/dcgm-gpu-stats-gpu201-23-l-34673507.out
DELETED
@@ -1,39 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34673507.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 1 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Fri Jul 12 11:48:45 2024 |
|
7 |
-
| End Time | Sat Jul 13 11:49:39 2024 |
|
8 |
-
| Total Execution Time (sec) | 86454.5 |
|
9 |
-
| No. of Processes | 1 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 252136 |
|
12 |
-
| Power Usage (Watts) | Avg: 69.7762, Max: 70.022, Min: 69.151 |
|
13 |
-
| Max GPU Memory Used (bytes) | 10104078336 |
|
14 |
-
| SM Clock (MHz) | Avg: 1157, Max: 1410, Min: 1155 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 2527521 |
|
34 |
-
| Avg SM Utilization (%) | 0 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
+----- Overall Health -------------+-----------------------------------------+
|
37 |
-
| Overall Health | Healthy |
|
38 |
-
+------------------------------------+-----------------------------------------+
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34676162/dcgm-gpu-stats-gpu201-23-l-34676162.out
DELETED
@@ -1,39 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34676162.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 3 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Sun Jul 14 07:57:08 2024 |
|
7 |
-
| End Time | Mon Jul 15 07:57:59 2024 |
|
8 |
-
| Total Execution Time (sec) | 86450.6 |
|
9 |
-
| No. of Processes | 1 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 249997 |
|
12 |
-
| Power Usage (Watts) | Avg: 82.8167, Max: 86.615, Min: 70.491 |
|
13 |
-
| Max GPU Memory Used (bytes) | 10104078336 |
|
14 |
-
| SM Clock (MHz) | Avg: 1352, Max: 1410, Min: 1080 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 3048225 |
|
34 |
-
| Avg SM Utilization (%) | 0 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
+----- Overall Health -------------+-----------------------------------------+
|
37 |
-
| Overall Health | Healthy |
|
38 |
-
+------------------------------------+-----------------------------------------+
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34691276/dcgm-gpu-stats-gpu201-09-l-34691276.out
DELETED
@@ -1,42 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34691276.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 0 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Tue Jul 16 08:21:43 2024 |
|
7 |
-
| End Time | Tue Jul 16 21:44:34 2024 |
|
8 |
-
| Total Execution Time (sec) | 48170.9 |
|
9 |
-
| No. of Processes | 2 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 222759 |
|
12 |
-
| Power Usage (Watts) | Avg: 61.4158, Max: 61.683, Min: 61.349 |
|
13 |
-
| Max GPU Memory Used (bytes) | 10806624256 |
|
14 |
-
| SM Clock (MHz) | Avg: 210, Max: 225, Min: 210 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 1958147 |
|
34 |
-
| Avg SM Utilization (%) | 1 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
| PID | 2068287 |
|
37 |
-
| Avg SM Utilization (%) | 0 |
|
38 |
-
| Avg Memory Utilization (%) | 0 |
|
39 |
-
+----- Overall Health -------------+-----------------------------------------+
|
40 |
-
| Overall Health | Healthy |
|
41 |
-
+------------------------------------+-----------------------------------------+
|
42 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34709014/dcgm-gpu-stats-gpu109-16-l-34709014.out
DELETED
@@ -1,39 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34709014.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 3 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Thu Jul 18 07:54:11 2024 |
|
7 |
-
| End Time | Fri Jul 19 07:55:07 2024 |
|
8 |
-
| Total Execution Time (sec) | 86456.3 |
|
9 |
-
| No. of Processes | 1 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 245376 |
|
12 |
-
| Power Usage (Watts) | Avg: 67.8347, Max: 68.156, Min: 67.563 |
|
13 |
-
| Max GPU Memory Used (bytes) | 10582228992 |
|
14 |
-
| SM Clock (MHz) | Avg: 1161, Max: 1410, Min: 1155 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 4005887 |
|
34 |
-
| Avg SM Utilization (%) | 0 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
+----- Overall Health -------------+-----------------------------------------+
|
37 |
-
| Overall Health | Healthy |
|
38 |
-
+------------------------------------+-----------------------------------------+
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34721198/dcgm-gpu-stats-gpu203-23-r-34721198.out
DELETED
@@ -1,57 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34721198.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 1 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Fri Jul 19 21:34:44 2024 |
|
7 |
-
| End Time | Sat Jul 20 00:01:06 2024 |
|
8 |
-
| Total Execution Time (sec) | 8782.23 |
|
9 |
-
| No. of Processes | 7 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 225540 |
|
12 |
-
| Power Usage (Watts) | Avg: 75.9496, Max: 87.541, Min: 65.792 |
|
13 |
-
| Max GPU Memory Used (bytes) | 13356761088 |
|
14 |
-
| SM Clock (MHz) | Avg: 210, Max: 210, Min: 210 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 866309 |
|
34 |
-
| Avg SM Utilization (%) | 0 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
| PID | 866955 |
|
37 |
-
| Avg SM Utilization (%) | 1 |
|
38 |
-
| Avg Memory Utilization (%) | 0 |
|
39 |
-
| PID | 868076 |
|
40 |
-
| Avg SM Utilization (%) | 0 |
|
41 |
-
| Avg Memory Utilization (%) | 0 |
|
42 |
-
| PID | 868638 |
|
43 |
-
| Avg SM Utilization (%) | 5 |
|
44 |
-
| Avg Memory Utilization (%) | 0 |
|
45 |
-
| PID | 869519 |
|
46 |
-
| Avg SM Utilization (%) | 0 |
|
47 |
-
| Avg Memory Utilization (%) | 0 |
|
48 |
-
| PID | 871043 |
|
49 |
-
| Avg SM Utilization (%) | 1 |
|
50 |
-
| Avg Memory Utilization (%) | 0 |
|
51 |
-
| PID | 871322 |
|
52 |
-
| Avg SM Utilization (%) | 0 |
|
53 |
-
| Avg Memory Utilization (%) | 0 |
|
54 |
-
+----- Overall Health -------------+-----------------------------------------+
|
55 |
-
| Overall Health | Healthy |
|
56 |
-
+------------------------------------+-----------------------------------------+
|
57 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34734121/dcgm-gpu-stats-gpu201-23-l-34734121.out
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34734121.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 3 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Tue Jul 23 11:47:49 2024 |
|
7 |
-
| End Time | Tue Jul 23 13:47:51 2024 |
|
8 |
-
| Total Execution Time (sec) | 7202.22 |
|
9 |
-
| No. of Processes | 0 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 226384 |
|
12 |
-
| Power Usage (Watts) | Avg: 62.6807, Max: 81.445, Min: 62.015 |
|
13 |
-
| Max GPU Memory Used (bytes) | 0 |
|
14 |
-
| SM Clock (MHz) | Avg: 220, Max: 1410, Min: 210 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+----- Overall Health -------------+-----------------------------------------+
|
33 |
-
| Overall Health | Healthy |
|
34 |
-
+------------------------------------+-----------------------------------------+
|
35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34738689/dcgm-gpu-stats-gpu201-16-r-34738689.out
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34738689.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 3 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Wed Jul 24 10:14:38 2024 |
|
7 |
-
| End Time | Wed Jul 24 11:45:33 2024 |
|
8 |
-
| Total Execution Time (sec) | 5454.69 |
|
9 |
-
| No. of Processes | 0 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 232516 |
|
12 |
-
| Power Usage (Watts) | Avg: 64.2532, Max: 64.329, Min: 63.938 |
|
13 |
-
| Max GPU Memory Used (bytes) | 0 |
|
14 |
-
| SM Clock (MHz) | Avg: 210, Max: 210, Min: 210 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+----- Overall Health -------------+-----------------------------------------+
|
33 |
-
| Overall Health | Healthy |
|
34 |
-
+------------------------------------+-----------------------------------------+
|
35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/dcgm/bash/34757693/dcgm-gpu-stats-gpu202-16-r-34757693.out
DELETED
@@ -1,42 +0,0 @@
|
|
1 |
-
Successfully retrieved statistics for job: 34757693.
|
2 |
-
+------------------------------------------------------------------------------+
|
3 |
-
| GPU ID: 2 |
|
4 |
-
+====================================+=========================================+
|
5 |
-
|----- Execution Stats ------------+-----------------------------------------|
|
6 |
-
| Start Time | Thu Jul 25 15:38:16 2024 |
|
7 |
-
| End Time | Thu Jul 25 17:08:59 2024 |
|
8 |
-
| Total Execution Time (sec) | 5442.54 |
|
9 |
-
| No. of Processes | 2 |
|
10 |
-
+----- Performance Stats ----------+-----------------------------------------+
|
11 |
-
| Energy Consumed (Joules) | 214029 |
|
12 |
-
| Power Usage (Watts) | Avg: 59.2012, Max: 67.659, Min: 59.026 |
|
13 |
-
| Max GPU Memory Used (bytes) | 7616856064 |
|
14 |
-
| SM Clock (MHz) | Avg: 243, Max: 1080, Min: 210 |
|
15 |
-
| Memory Clock (MHz) | Avg: 1593, Max: 1593, Min: 1593 |
|
16 |
-
| SM Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
17 |
-
| Memory Utilization (%) | Avg: 0, Max: 0, Min: 0 |
|
18 |
-
| PCIe Rx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
19 |
-
| PCIe Tx Bandwidth (megabytes) | Avg: N/A, Max: N/A, Min: N/A |
|
20 |
-
+----- Event Stats ----------------+-----------------------------------------+
|
21 |
-
| Single Bit ECC Errors | 0 |
|
22 |
-
| Double Bit ECC Errors | 0 |
|
23 |
-
| PCIe Replay Warnings | 0 |
|
24 |
-
| Critical XID Errors | 0 |
|
25 |
-
+----- Slowdown Stats -------------+-----------------------------------------+
|
26 |
-
| Due to - Power (%) | 0 |
|
27 |
-
| - Thermal (%) | 0 |
|
28 |
-
| - Reliability (%) | Not Supported |
|
29 |
-
| - Board Limit (%) | Not Supported |
|
30 |
-
| - Low Utilization (%) | Not Supported |
|
31 |
-
| - Sync Boost (%) | 0 |
|
32 |
-
+-- Compute Process Utilization ---+-----------------------------------------+
|
33 |
-
| PID | 1095606 |
|
34 |
-
| Avg SM Utilization (%) | 3 |
|
35 |
-
| Avg Memory Utilization (%) | 0 |
|
36 |
-
| PID | 1096190 |
|
37 |
-
| Avg SM Utilization (%) | 14 |
|
38 |
-
| Avg Memory Utilization (%) | 2 |
|
39 |
-
+----- Overall Health -------------+-----------------------------------------+
|
40 |
-
| Overall Health | Healthy |
|
41 |
-
+------------------------------------+-----------------------------------------+
|
42 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/demo_v2.py
DELETED
@@ -1,648 +0,0 @@
|
|
1 |
-
# python demo_v2.py --cfg-path eval_configs/minigptv2_eval.yaml --gpu-id 0
|
2 |
-
|
3 |
-
import argparse
|
4 |
-
import os
|
5 |
-
import random
|
6 |
-
from collections import defaultdict
|
7 |
-
|
8 |
-
import cv2
|
9 |
-
import re
|
10 |
-
|
11 |
-
import numpy as np
|
12 |
-
from PIL import Image
|
13 |
-
import torch
|
14 |
-
import html
|
15 |
-
import gradio as gr
|
16 |
-
|
17 |
-
import torchvision.transforms as T
|
18 |
-
import torch.backends.cudnn as cudnn
|
19 |
-
|
20 |
-
from minigpt4.common.config import Config
|
21 |
-
|
22 |
-
from minigpt4.common.registry import registry
|
23 |
-
from minigpt4.conversation.conversation import Conversation, SeparatorStyle, Chat
|
24 |
-
|
25 |
-
# imports modules for registration
|
26 |
-
from minigpt4.datasets.builders import *
|
27 |
-
from minigpt4.models import *
|
28 |
-
from minigpt4.processors import *
|
29 |
-
from minigpt4.runners import *
|
30 |
-
from minigpt4.tasks import *
|
31 |
-
|
32 |
-
|
33 |
-
def parse_args():
|
34 |
-
parser = argparse.ArgumentParser(description="Demo")
|
35 |
-
parser.add_argument("--cfg-path", default='eval_configs/minigptv2_eval.yaml',
|
36 |
-
help="path to configuration file.")
|
37 |
-
parser.add_argument("--gpu-id", type=int, default=0, help="specify the gpu to load the model.")
|
38 |
-
parser.add_argument(
|
39 |
-
"--options",
|
40 |
-
nargs="+",
|
41 |
-
help="override some settings in the used config, the key-value pair "
|
42 |
-
"in xxx=yyy format will be merged into config file (deprecate), "
|
43 |
-
"change to --cfg-options instead.",
|
44 |
-
)
|
45 |
-
args = parser.parse_args()
|
46 |
-
return args
|
47 |
-
|
48 |
-
|
49 |
-
random.seed(42)
|
50 |
-
np.random.seed(42)
|
51 |
-
torch.manual_seed(42)
|
52 |
-
|
53 |
-
cudnn.benchmark = False
|
54 |
-
cudnn.deterministic = True
|
55 |
-
|
56 |
-
print('Initializing Chat')
|
57 |
-
args = parse_args()
|
58 |
-
cfg = Config(args)
|
59 |
-
|
60 |
-
device = 'cuda:{}'.format(args.gpu_id)
|
61 |
-
|
62 |
-
model_config = cfg.model_cfg
|
63 |
-
model_config.device_8bit = args.gpu_id
|
64 |
-
model_cls = registry.get_model_class(model_config.arch)
|
65 |
-
model = model_cls.from_config(model_config).to(device)
|
66 |
-
bounding_box_size = 100
|
67 |
-
|
68 |
-
vis_processor_cfg = cfg.datasets_cfg.cc_sbu_align.vis_processor.train
|
69 |
-
vis_processor = registry.get_processor_class(vis_processor_cfg.name).from_config(vis_processor_cfg)
|
70 |
-
|
71 |
-
model = model.eval()
|
72 |
-
|
73 |
-
CONV_VISION = Conversation(
|
74 |
-
system="",
|
75 |
-
roles=(r"<s>[INST] ", r" [/INST]"),
|
76 |
-
messages=[],
|
77 |
-
offset=2,
|
78 |
-
sep_style=SeparatorStyle.SINGLE,
|
79 |
-
sep="",
|
80 |
-
)
|
81 |
-
|
82 |
-
|
83 |
-
def extract_substrings(string):
|
84 |
-
# first check if there is no-finished bracket
|
85 |
-
index = string.rfind('}')
|
86 |
-
if index != -1:
|
87 |
-
string = string[:index + 1]
|
88 |
-
|
89 |
-
pattern = r'<p>(.*?)\}(?!<)'
|
90 |
-
matches = re.findall(pattern, string)
|
91 |
-
substrings = [match for match in matches]
|
92 |
-
|
93 |
-
return substrings
|
94 |
-
|
95 |
-
|
96 |
-
def is_overlapping(rect1, rect2):
|
97 |
-
x1, y1, x2, y2 = rect1
|
98 |
-
x3, y3, x4, y4 = rect2
|
99 |
-
return not (x2 < x3 or x1 > x4 or y2 < y3 or y1 > y4)
|
100 |
-
|
101 |
-
|
102 |
-
def computeIoU(bbox1, bbox2):
|
103 |
-
x1, y1, x2, y2 = bbox1
|
104 |
-
x3, y3, x4, y4 = bbox2
|
105 |
-
intersection_x1 = max(x1, x3)
|
106 |
-
intersection_y1 = max(y1, y3)
|
107 |
-
intersection_x2 = min(x2, x4)
|
108 |
-
intersection_y2 = min(y2, y4)
|
109 |
-
intersection_area = max(0, intersection_x2 - intersection_x1 + 1) * max(0, intersection_y2 - intersection_y1 + 1)
|
110 |
-
bbox1_area = (x2 - x1 + 1) * (y2 - y1 + 1)
|
111 |
-
bbox2_area = (x4 - x3 + 1) * (y4 - y3 + 1)
|
112 |
-
union_area = bbox1_area + bbox2_area - intersection_area
|
113 |
-
iou = intersection_area / union_area
|
114 |
-
return iou
|
115 |
-
|
116 |
-
|
117 |
-
def save_tmp_img(visual_img):
|
118 |
-
file_name = "".join([str(random.randint(0, 9)) for _ in range(5)]) + ".jpg"
|
119 |
-
file_path = "/tmp/gradio" + file_name
|
120 |
-
visual_img.save(file_path)
|
121 |
-
return file_path
|
122 |
-
|
123 |
-
|
124 |
-
def mask2bbox(mask):
|
125 |
-
if mask is None:
|
126 |
-
return ''
|
127 |
-
mask = mask.resize([100, 100], resample=Image.NEAREST)
|
128 |
-
mask = np.array(mask)[:, :, 0]
|
129 |
-
|
130 |
-
rows = np.any(mask, axis=1)
|
131 |
-
cols = np.any(mask, axis=0)
|
132 |
-
|
133 |
-
if rows.sum():
|
134 |
-
# Get the top, bottom, left, and right boundaries
|
135 |
-
rmin, rmax = np.where(rows)[0][[0, -1]]
|
136 |
-
cmin, cmax = np.where(cols)[0][[0, -1]]
|
137 |
-
bbox = '{{<{}><{}><{}><{}>}}'.format(cmin, rmin, cmax, rmax)
|
138 |
-
else:
|
139 |
-
bbox = ''
|
140 |
-
|
141 |
-
return bbox
|
142 |
-
|
143 |
-
|
144 |
-
def escape_markdown(text):
|
145 |
-
# List of Markdown special characters that need to be escaped
|
146 |
-
md_chars = ['<', '>']
|
147 |
-
|
148 |
-
# Escape each special character
|
149 |
-
for char in md_chars:
|
150 |
-
text = text.replace(char, '\\' + char)
|
151 |
-
|
152 |
-
return text
|
153 |
-
|
154 |
-
|
155 |
-
def reverse_escape(text):
|
156 |
-
md_chars = ['\\<', '\\>']
|
157 |
-
|
158 |
-
for char in md_chars:
|
159 |
-
text = text.replace(char, char[1:])
|
160 |
-
|
161 |
-
return text
|
162 |
-
|
163 |
-
|
164 |
-
colors = [
|
165 |
-
(255, 0, 0),
|
166 |
-
(0, 255, 0),
|
167 |
-
(0, 0, 255),
|
168 |
-
(210, 210, 0),
|
169 |
-
(255, 0, 255),
|
170 |
-
(0, 255, 255),
|
171 |
-
(114, 128, 250),
|
172 |
-
(0, 165, 255),
|
173 |
-
(0, 128, 0),
|
174 |
-
(144, 238, 144),
|
175 |
-
(238, 238, 175),
|
176 |
-
(255, 191, 0),
|
177 |
-
(0, 128, 0),
|
178 |
-
(226, 43, 138),
|
179 |
-
(255, 0, 255),
|
180 |
-
(0, 215, 255),
|
181 |
-
]
|
182 |
-
|
183 |
-
color_map = {
|
184 |
-
f"{color_id}": f"#{hex(color[2])[2:].zfill(2)}{hex(color[1])[2:].zfill(2)}{hex(color[0])[2:].zfill(2)}" for
|
185 |
-
color_id, color in enumerate(colors)
|
186 |
-
}
|
187 |
-
|
188 |
-
used_colors = colors
|
189 |
-
|
190 |
-
|
191 |
-
def visualize_all_bbox_together(image, generation):
|
192 |
-
if image is None:
|
193 |
-
return None, ''
|
194 |
-
|
195 |
-
generation = html.unescape(generation)
|
196 |
-
print('gen begin', generation)
|
197 |
-
image_width, image_height = image.size
|
198 |
-
image = image.resize([500, int(500 / image_width * image_height)])
|
199 |
-
image_width, image_height = image.size
|
200 |
-
|
201 |
-
string_list = extract_substrings(generation)
|
202 |
-
if string_list: # it is grounding or detection
|
203 |
-
mode = 'all'
|
204 |
-
entities = defaultdict(list)
|
205 |
-
i = 0
|
206 |
-
j = 0
|
207 |
-
for string in string_list:
|
208 |
-
try:
|
209 |
-
obj, string = string.split('</p>')
|
210 |
-
except ValueError:
|
211 |
-
print('wrong string: ', string)
|
212 |
-
continue
|
213 |
-
bbox_list = string.split('<delim>')
|
214 |
-
flag = False
|
215 |
-
for bbox_string in bbox_list:
|
216 |
-
integers = re.findall(r'-?\d+', bbox_string)
|
217 |
-
if len(integers) == 4:
|
218 |
-
x0, y0, x1, y1 = int(integers[0]), int(integers[1]), int(integers[2]), int(integers[3])
|
219 |
-
left = x0 / bounding_box_size * image_width
|
220 |
-
bottom = y0 / bounding_box_size * image_height
|
221 |
-
right = x1 / bounding_box_size * image_width
|
222 |
-
top = y1 / bounding_box_size * image_height
|
223 |
-
|
224 |
-
entities[obj].append([left, bottom, right, top])
|
225 |
-
|
226 |
-
j += 1
|
227 |
-
flag = True
|
228 |
-
if flag:
|
229 |
-
i += 1
|
230 |
-
else:
|
231 |
-
integers = re.findall(r'-?\d+', generation)
|
232 |
-
|
233 |
-
if len(integers) == 4: # it is refer
|
234 |
-
mode = 'single'
|
235 |
-
|
236 |
-
entities = list()
|
237 |
-
x0, y0, x1, y1 = int(integers[0]), int(integers[1]), int(integers[2]), int(integers[3])
|
238 |
-
left = x0 / bounding_box_size * image_width
|
239 |
-
bottom = y0 / bounding_box_size * image_height
|
240 |
-
right = x1 / bounding_box_size * image_width
|
241 |
-
top = y1 / bounding_box_size * image_height
|
242 |
-
entities.append([left, bottom, right, top])
|
243 |
-
else:
|
244 |
-
# don't detect any valid bbox to visualize
|
245 |
-
return None, ''
|
246 |
-
|
247 |
-
if len(entities) == 0:
|
248 |
-
return None, ''
|
249 |
-
|
250 |
-
if isinstance(image, Image.Image):
|
251 |
-
image_h = image.height
|
252 |
-
image_w = image.width
|
253 |
-
image = np.array(image)
|
254 |
-
|
255 |
-
elif isinstance(image, str):
|
256 |
-
if os.path.exists(image):
|
257 |
-
pil_img = Image.open(image).convert("RGB")
|
258 |
-
image = np.array(pil_img)[:, :, [2, 1, 0]]
|
259 |
-
image_h = pil_img.height
|
260 |
-
image_w = pil_img.width
|
261 |
-
else:
|
262 |
-
raise ValueError(f"invaild image path, {image}")
|
263 |
-
elif isinstance(image, torch.Tensor):
|
264 |
-
|
265 |
-
image_tensor = image.cpu()
|
266 |
-
reverse_norm_mean = torch.tensor([0.48145466, 0.4578275, 0.40821073])[:, None, None]
|
267 |
-
reverse_norm_std = torch.tensor([0.26862954, 0.26130258, 0.27577711])[:, None, None]
|
268 |
-
image_tensor = image_tensor * reverse_norm_std + reverse_norm_mean
|
269 |
-
pil_img = T.ToPILImage()(image_tensor)
|
270 |
-
image_h = pil_img.height
|
271 |
-
image_w = pil_img.width
|
272 |
-
image = np.array(pil_img)[:, :, [2, 1, 0]]
|
273 |
-
else:
|
274 |
-
raise ValueError(f"invaild image format, {type(image)} for {image}")
|
275 |
-
|
276 |
-
indices = list(range(len(entities)))
|
277 |
-
|
278 |
-
new_image = image.copy()
|
279 |
-
|
280 |
-
previous_bboxes = []
|
281 |
-
# size of text
|
282 |
-
text_size = 0.5
|
283 |
-
# thickness of text
|
284 |
-
text_line = 1 # int(max(1 * min(image_h, image_w) / 512, 1))
|
285 |
-
box_line = 2
|
286 |
-
(c_width, text_height), _ = cv2.getTextSize("F", cv2.FONT_HERSHEY_COMPLEX, text_size, text_line)
|
287 |
-
base_height = int(text_height * 0.675)
|
288 |
-
text_offset_original = text_height - base_height
|
289 |
-
text_spaces = 2
|
290 |
-
|
291 |
-
# num_bboxes = sum(len(x[-1]) for x in entities)
|
292 |
-
used_colors = colors # random.sample(colors, k=num_bboxes)
|
293 |
-
|
294 |
-
color_id = -1
|
295 |
-
for entity_idx, entity_name in enumerate(entities):
|
296 |
-
if mode == 'single' or mode == 'identify':
|
297 |
-
bboxes = entity_name
|
298 |
-
bboxes = [bboxes]
|
299 |
-
else:
|
300 |
-
bboxes = entities[entity_name]
|
301 |
-
color_id += 1
|
302 |
-
for bbox_id, (x1_norm, y1_norm, x2_norm, y2_norm) in enumerate(bboxes):
|
303 |
-
skip_flag = False
|
304 |
-
orig_x1, orig_y1, orig_x2, orig_y2 = int(x1_norm), int(y1_norm), int(x2_norm), int(y2_norm)
|
305 |
-
|
306 |
-
color = used_colors[entity_idx % len(used_colors)] # tuple(np.random.randint(0, 255, size=3).tolist())
|
307 |
-
new_image = cv2.rectangle(new_image, (orig_x1, orig_y1), (orig_x2, orig_y2), color, box_line)
|
308 |
-
|
309 |
-
if mode == 'all':
|
310 |
-
l_o, r_o = box_line // 2 + box_line % 2, box_line // 2 + box_line % 2 + 1
|
311 |
-
|
312 |
-
x1 = orig_x1 - l_o
|
313 |
-
y1 = orig_y1 - l_o
|
314 |
-
|
315 |
-
if y1 < text_height + text_offset_original + 2 * text_spaces:
|
316 |
-
y1 = orig_y1 + r_o + text_height + text_offset_original + 2 * text_spaces
|
317 |
-
x1 = orig_x1 + r_o
|
318 |
-
|
319 |
-
# add text background
|
320 |
-
(text_width, text_height), _ = cv2.getTextSize(f" {entity_name}", cv2.FONT_HERSHEY_COMPLEX, text_size,
|
321 |
-
text_line)
|
322 |
-
text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2 = x1, y1 - (
|
323 |
-
text_height + text_offset_original + 2 * text_spaces), x1 + text_width, y1
|
324 |
-
|
325 |
-
for prev_bbox in previous_bboxes:
|
326 |
-
if computeIoU((text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), prev_bbox['bbox']) > 0.95 and \
|
327 |
-
prev_bbox['phrase'] == entity_name:
|
328 |
-
skip_flag = True
|
329 |
-
break
|
330 |
-
while is_overlapping((text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), prev_bbox['bbox']):
|
331 |
-
text_bg_y1 += (text_height + text_offset_original + 2 * text_spaces)
|
332 |
-
text_bg_y2 += (text_height + text_offset_original + 2 * text_spaces)
|
333 |
-
y1 += (text_height + text_offset_original + 2 * text_spaces)
|
334 |
-
|
335 |
-
if text_bg_y2 >= image_h:
|
336 |
-
text_bg_y1 = max(0, image_h - (text_height + text_offset_original + 2 * text_spaces))
|
337 |
-
text_bg_y2 = image_h
|
338 |
-
y1 = image_h
|
339 |
-
break
|
340 |
-
if not skip_flag:
|
341 |
-
alpha = 0.5
|
342 |
-
for i in range(text_bg_y1, text_bg_y2):
|
343 |
-
for j in range(text_bg_x1, text_bg_x2):
|
344 |
-
if i < image_h and j < image_w:
|
345 |
-
if j < text_bg_x1 + 1.35 * c_width:
|
346 |
-
# original color
|
347 |
-
bg_color = color
|
348 |
-
else:
|
349 |
-
# white
|
350 |
-
bg_color = [255, 255, 255]
|
351 |
-
new_image[i, j] = (alpha * new_image[i, j] + (1 - alpha) * np.array(bg_color)).astype(
|
352 |
-
np.uint8)
|
353 |
-
|
354 |
-
cv2.putText(
|
355 |
-
new_image, f" {entity_name}", (x1, y1 - text_offset_original - 1 * text_spaces),
|
356 |
-
cv2.FONT_HERSHEY_COMPLEX, text_size, (0, 0, 0), text_line, cv2.LINE_AA
|
357 |
-
)
|
358 |
-
|
359 |
-
previous_bboxes.append(
|
360 |
-
{'bbox': (text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), 'phrase': entity_name})
|
361 |
-
|
362 |
-
if mode == 'all':
|
363 |
-
def color_iterator(colors):
|
364 |
-
while True:
|
365 |
-
for color in colors:
|
366 |
-
yield color
|
367 |
-
|
368 |
-
color_gen = color_iterator(colors)
|
369 |
-
|
370 |
-
# Add colors to phrases and remove <p></p>
|
371 |
-
def colored_phrases(match):
|
372 |
-
phrase = match.group(1)
|
373 |
-
color = next(color_gen)
|
374 |
-
return f'<span style="color:rgb{color}">{phrase}</span>'
|
375 |
-
|
376 |
-
generation = re.sub(r'{<\d+><\d+><\d+><\d+>}|<delim>', '', generation)
|
377 |
-
generation_colored = re.sub(r'<p>(.*?)</p>', colored_phrases, generation)
|
378 |
-
else:
|
379 |
-
generation_colored = ''
|
380 |
-
|
381 |
-
pil_image = Image.fromarray(new_image)
|
382 |
-
return pil_image, generation_colored
|
383 |
-
|
384 |
-
|
385 |
-
def gradio_reset(chat_state, img_list):
|
386 |
-
if chat_state is not None:
|
387 |
-
chat_state.messages = []
|
388 |
-
if img_list is not None:
|
389 |
-
img_list = []
|
390 |
-
return None, gr.update(value=None, interactive=True), gr.update(placeholder='Upload your image and chat',
|
391 |
-
interactive=True), chat_state, img_list
|
392 |
-
|
393 |
-
|
394 |
-
def image_upload_trigger(upload_flag, replace_flag, img_list):
|
395 |
-
# set the upload flag to true when receive a new image.
|
396 |
-
# if there is an old image (and old conversation), set the replace flag to true to reset the conv later.
|
397 |
-
upload_flag = 1
|
398 |
-
if img_list:
|
399 |
-
replace_flag = 1
|
400 |
-
return upload_flag, replace_flag
|
401 |
-
|
402 |
-
|
403 |
-
def example_trigger(text_input, image, upload_flag, replace_flag, img_list):
|
404 |
-
# set the upload flag to true when receive a new image.
|
405 |
-
# if there is an old image (and old conversation), set the replace flag to true to reset the conv later.
|
406 |
-
upload_flag = 1
|
407 |
-
if img_list or replace_flag == 1:
|
408 |
-
replace_flag = 1
|
409 |
-
|
410 |
-
return upload_flag, replace_flag
|
411 |
-
|
412 |
-
|
413 |
-
def gradio_ask(user_message, chatbot, chat_state, gr_img, img_list, upload_flag, replace_flag):
|
414 |
-
if len(user_message) == 0:
|
415 |
-
text_box_show = 'Input should not be empty!'
|
416 |
-
else:
|
417 |
-
text_box_show = ''
|
418 |
-
|
419 |
-
if isinstance(gr_img, dict):
|
420 |
-
gr_img, mask = gr_img['image'], gr_img['mask']
|
421 |
-
else:
|
422 |
-
mask = None
|
423 |
-
|
424 |
-
if '[identify]' in user_message:
|
425 |
-
# check if user provide bbox in the text input
|
426 |
-
integers = re.findall(r'-?\d+', user_message)
|
427 |
-
if len(integers) != 4: # no bbox in text
|
428 |
-
bbox = mask2bbox(mask)
|
429 |
-
user_message = user_message + bbox
|
430 |
-
|
431 |
-
if chat_state is None:
|
432 |
-
chat_state = CONV_VISION.copy()
|
433 |
-
|
434 |
-
if upload_flag:
|
435 |
-
if replace_flag:
|
436 |
-
chat_state = CONV_VISION.copy() # new image, reset everything
|
437 |
-
replace_flag = 0
|
438 |
-
chatbot = []
|
439 |
-
img_list = []
|
440 |
-
llm_message = chat.upload_img(gr_img, chat_state, img_list)
|
441 |
-
upload_flag = 0
|
442 |
-
|
443 |
-
chat.ask(user_message, chat_state)
|
444 |
-
|
445 |
-
chatbot = chatbot + [[user_message, None]]
|
446 |
-
|
447 |
-
if '[identify]' in user_message:
|
448 |
-
visual_img, _ = visualize_all_bbox_together(gr_img, user_message)
|
449 |
-
if visual_img is not None:
|
450 |
-
file_path = save_tmp_img(visual_img)
|
451 |
-
chatbot = chatbot + [[(file_path,), None]]
|
452 |
-
|
453 |
-
return text_box_show, chatbot, chat_state, img_list, upload_flag, replace_flag
|
454 |
-
|
455 |
-
|
456 |
-
def gradio_answer(chatbot, chat_state, img_list, temperature):
|
457 |
-
llm_message = chat.answer(conv=chat_state,
|
458 |
-
img_list=img_list,
|
459 |
-
temperature=temperature,
|
460 |
-
max_new_tokens=500,
|
461 |
-
max_length=2000)[0]
|
462 |
-
chatbot[-1][1] = llm_message
|
463 |
-
return chatbot, chat_state
|
464 |
-
|
465 |
-
|
466 |
-
def gradio_stream_answer(chatbot, chat_state, img_list, temperature):
|
467 |
-
if len(img_list) > 0:
|
468 |
-
if not isinstance(img_list[0], torch.Tensor):
|
469 |
-
chat.encode_img(img_list)
|
470 |
-
streamer = chat.stream_answer(conv=chat_state,
|
471 |
-
img_list=img_list,
|
472 |
-
temperature=temperature,
|
473 |
-
max_new_tokens=500,
|
474 |
-
max_length=2000)
|
475 |
-
output = ''
|
476 |
-
for new_output in streamer:
|
477 |
-
escapped = escape_markdown(new_output)
|
478 |
-
output += escapped
|
479 |
-
chatbot[-1][1] = output
|
480 |
-
yield chatbot, chat_state
|
481 |
-
chat_state.messages[-1][1] = '</s>'
|
482 |
-
return chatbot, chat_state
|
483 |
-
|
484 |
-
|
485 |
-
def gradio_visualize(chatbot, gr_img):
|
486 |
-
if isinstance(gr_img, dict):
|
487 |
-
gr_img, mask = gr_img['image'], gr_img['mask']
|
488 |
-
|
489 |
-
unescaped = reverse_escape(chatbot[-1][1])
|
490 |
-
visual_img, generation_color = visualize_all_bbox_together(gr_img, unescaped)
|
491 |
-
if visual_img is not None:
|
492 |
-
if len(generation_color):
|
493 |
-
chatbot[-1][1] = generation_color
|
494 |
-
file_path = save_tmp_img(visual_img)
|
495 |
-
chatbot = chatbot + [[None, (file_path,)]]
|
496 |
-
|
497 |
-
return chatbot
|
498 |
-
|
499 |
-
|
500 |
-
def gradio_taskselect(idx):
|
501 |
-
prompt_list = [
|
502 |
-
'',
|
503 |
-
'[grounding] describe this image in detail',
|
504 |
-
'[refer] ',
|
505 |
-
'[detection] ',
|
506 |
-
'[identify] what is this ',
|
507 |
-
'[vqa] '
|
508 |
-
]
|
509 |
-
instruct_list = [
|
510 |
-
'**Hint:** Type in whatever you want',
|
511 |
-
'**Hint:** Send the command to generate a grounded image description',
|
512 |
-
'**Hint:** Type in a phrase about an object in the image and send the command',
|
513 |
-
'**Hint:** Type in a caption or phrase, and see object locations in the image',
|
514 |
-
'**Hint:** Draw a bounding box on the uploaded image then send the command. Click the "clear" botton on the top right of the image before redraw',
|
515 |
-
'**Hint:** Send a question to get a short answer',
|
516 |
-
]
|
517 |
-
return prompt_list[idx], instruct_list[idx]
|
518 |
-
|
519 |
-
|
520 |
-
|
521 |
-
|
522 |
-
chat = Chat(model, vis_processor, device=device)
|
523 |
-
|
524 |
-
title = """<h1 align="center">MiniGPT-Med Demo</h1>"""
|
525 |
-
description = 'Welcome to Our MiniGPT-Med Chatbot Demo!'
|
526 |
-
# article = """<p><a href='https://minigpt-v2.github.io'><img src='https://img.shields.io/badge/Project-Page-Green'></a></p><p><a href='https://github.com/Vision-CAIR/MiniGPT-4/blob/main/MiniGPTv2.pdf'><img src='https://img.shields.io/badge/Paper-PDF-red'></a></p><p><a href='https://github.com/Vision-CAIR/MiniGPT-4'><img src='https://img.shields.io/badge/GitHub-Repo-blue'></a></p><p><a href='https://www.youtube.com/watch?v=atFCwV2hSY4'><img src='https://img.shields.io/badge/YouTube-Video-red'></a></p>"""
|
527 |
-
article = """<p><a href='https://minigpt-med.github.io/'><img src='https://img.shields.io/badge/Project-Page-Green'></a></p>"""
|
528 |
-
|
529 |
-
introduction = '''
|
530 |
-
For Abilities Involving Visual Grounding:
|
531 |
-
1. Grounding: CLICK **Send** to generate a grounded image description.
|
532 |
-
2. Refer: Input a referring object and CLICK **Send**.
|
533 |
-
3. Detection: Write a caption or phrase, and CLICK **Send**.
|
534 |
-
4. Identify: Draw the bounding box on the uploaded image window and CLICK **Send** to generate the bounding box. (CLICK "clear" button before re-drawing next time).
|
535 |
-
5. VQA: Input a visual question and CLICK **Send**.
|
536 |
-
6. No Tag: Input whatever you want and CLICK **Send** without any tagging
|
537 |
-
|
538 |
-
You can also simply chat in free form!
|
539 |
-
'''
|
540 |
-
|
541 |
-
text_input = gr.Textbox(placeholder='Upload your image and chat', interactive=True, show_label=False, container=False,
|
542 |
-
scale=8)
|
543 |
-
with gr.Blocks() as demo:
|
544 |
-
gr.Markdown(title)
|
545 |
-
# gr.Markdown(description)
|
546 |
-
gr.Markdown(article)
|
547 |
-
|
548 |
-
with gr.Row():
|
549 |
-
with gr.Column(scale=0.5):
|
550 |
-
image = gr.Image(type="pil", tool='sketch', brush_radius=20)
|
551 |
-
|
552 |
-
temperature = gr.Slider(
|
553 |
-
minimum=0.1,
|
554 |
-
maximum=1.5,
|
555 |
-
value=0.6,
|
556 |
-
step=0.1,
|
557 |
-
interactive=True,
|
558 |
-
label="Temperature",
|
559 |
-
)
|
560 |
-
|
561 |
-
clear = gr.Button("Restart")
|
562 |
-
|
563 |
-
gr.Markdown(introduction)
|
564 |
-
|
565 |
-
with gr.Column():
|
566 |
-
chat_state = gr.State(value=None)
|
567 |
-
img_list = gr.State(value=[])
|
568 |
-
chatbot = gr.Chatbot(label='MiniGPT-Med')
|
569 |
-
|
570 |
-
dataset = gr.Dataset(
|
571 |
-
components=[gr.Textbox(visible=False)],
|
572 |
-
samples=[['No Tag'], ['Grounding'], ['Refer'], ['Detection'], ['Identify'], ['VQA']],
|
573 |
-
type="index",
|
574 |
-
label='Task Shortcuts',
|
575 |
-
)
|
576 |
-
task_inst = gr.Markdown('**Hint:** Upload your image and chat')
|
577 |
-
with gr.Row():
|
578 |
-
text_input.render()
|
579 |
-
send = gr.Button("Send", variant='primary', size='sm', scale=1)
|
580 |
-
|
581 |
-
upload_flag = gr.State(value=0)
|
582 |
-
replace_flag = gr.State(value=0)
|
583 |
-
image.upload(image_upload_trigger, [upload_flag, replace_flag, img_list], [upload_flag, replace_flag])
|
584 |
-
# [29, 44, 42, 56]
|
585 |
-
with gr.Row():
|
586 |
-
with gr.Column():
|
587 |
-
gr.Examples(examples=[
|
588 |
-
["Med_examples_v2/xmlab149/source.jpg", "[identify] what is this {<56><16><84><58>}", upload_flag,
|
589 |
-
replace_flag, img_list],
|
590 |
-
["Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.1495.1517874291.249176.jpg", "[detection] pneumonia", upload_flag, replace_flag, img_list],
|
591 |
-
["Med_examples_v2/1.2.840.113654.2.55.48339325922382839066544590341580673064.png", "[refer] the nodule in the left lung", upload_flag, replace_flag,
|
592 |
-
img_list],
|
593 |
-
["Med_examples_v2/xmlab589/source.jpg", "[grounding] describe this image in detail", upload_flag, replace_flag, img_list],
|
594 |
-
], inputs=[image, text_input, upload_flag, replace_flag, img_list], fn=example_trigger,
|
595 |
-
outputs=[upload_flag, replace_flag])
|
596 |
-
with gr.Column():
|
597 |
-
gr.Examples(examples=[
|
598 |
-
["Med_examples_v2/synpic50958.jpg", "[vqa] What does the small white lesions in the aorta mean?",
|
599 |
-
upload_flag, replace_flag, img_list],
|
600 |
-
["Med_examples_v2/5f4e8079-8225a5d2-1b0c3c46-4394a094-f285db0e.jpg", "Please provide a detailed description of the picture", upload_flag, replace_flag, img_list],
|
601 |
-
["Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.16254.1517874395.786150.jpg", "Diagnose this image", upload_flag, replace_flag, img_list],
|
602 |
-
["Med_examples_v2/synpic58547.jpg", "Could you describe the contents of this image for me?", upload_flag,
|
603 |
-
replace_flag, img_list],
|
604 |
-
], inputs=[image, text_input, upload_flag, replace_flag, img_list], fn=example_trigger,
|
605 |
-
outputs=[upload_flag, replace_flag])
|
606 |
-
|
607 |
-
dataset.click(
|
608 |
-
gradio_taskselect,
|
609 |
-
inputs=[dataset],
|
610 |
-
outputs=[text_input, task_inst],
|
611 |
-
show_progress="hidden",
|
612 |
-
postprocess=False,
|
613 |
-
queue=False,
|
614 |
-
)
|
615 |
-
|
616 |
-
text_input.submit(
|
617 |
-
gradio_ask,
|
618 |
-
[text_input, chatbot, chat_state, image, img_list, upload_flag, replace_flag],
|
619 |
-
[text_input, chatbot, chat_state, img_list, upload_flag, replace_flag], queue=False
|
620 |
-
).success(
|
621 |
-
gradio_stream_answer,
|
622 |
-
[chatbot, chat_state, img_list, temperature],
|
623 |
-
[chatbot, chat_state]
|
624 |
-
).success(
|
625 |
-
gradio_visualize,
|
626 |
-
[chatbot, image],
|
627 |
-
[chatbot],
|
628 |
-
queue=False,
|
629 |
-
)
|
630 |
-
|
631 |
-
send.click(
|
632 |
-
gradio_ask,
|
633 |
-
[text_input, chatbot, chat_state, image, img_list, upload_flag, replace_flag],
|
634 |
-
[text_input, chatbot, chat_state, img_list, upload_flag, replace_flag], queue=False
|
635 |
-
).success(
|
636 |
-
gradio_stream_answer,
|
637 |
-
[chatbot, chat_state, img_list, temperature],
|
638 |
-
[chatbot, chat_state]
|
639 |
-
).success(
|
640 |
-
gradio_visualize,
|
641 |
-
[chatbot, image],
|
642 |
-
[chatbot],
|
643 |
-
queue=False,
|
644 |
-
)
|
645 |
-
|
646 |
-
clear.click(gradio_reset, [chat_state, img_list], [chatbot, image, text_input, chat_state, img_list], queue=False)
|
647 |
-
|
648 |
-
demo.launch(share=True, enable_queue=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/environment.yml
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
name: miniGPT-Med
|
2 |
-
channels:
|
3 |
-
- pytorch
|
4 |
-
- defaults
|
5 |
-
- anaconda
|
6 |
-
dependencies:
|
7 |
-
- python=3.9
|
8 |
-
- cudatoolkit
|
9 |
-
- pip
|
10 |
-
- pip:
|
11 |
-
- torch==2.0.0
|
12 |
-
- torchaudio
|
13 |
-
- torchvision
|
14 |
-
- huggingface-hub==0.18.0
|
15 |
-
- matplotlib==3.7.0
|
16 |
-
- psutil==5.9.4
|
17 |
-
- iopath
|
18 |
-
- pyyaml==6.0
|
19 |
-
- regex==2022.10.31
|
20 |
-
- tokenizers==0.13.2
|
21 |
-
- tqdm==4.64.1
|
22 |
-
- transformers==4.30.0
|
23 |
-
- timm==0.6.13
|
24 |
-
- webdataset==0.2.48
|
25 |
-
- omegaconf==2.3.0
|
26 |
-
- opencv-python==4.7.0.72
|
27 |
-
- decord==0.6.0
|
28 |
-
- peft==0.2.0
|
29 |
-
- sentence-transformers
|
30 |
-
- gradio==3.47.1
|
31 |
-
- accelerate==0.20.3
|
32 |
-
- bitsandbytes==0.37.0
|
33 |
-
- scikit-image
|
34 |
-
- visual-genome
|
35 |
-
- wandb
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/eval_configs/minigptv2_benchmark_evaluation.yaml
DELETED
@@ -1,69 +0,0 @@
|
|
1 |
-
model:
|
2 |
-
arch: minigpt_v2
|
3 |
-
model_type: pretrain
|
4 |
-
max_txt_len: 500
|
5 |
-
end_sym: "</s>"
|
6 |
-
low_resource: False
|
7 |
-
prompt_template: '[INST] {} [/INST]'
|
8 |
-
llama_model: "/ibex/project/c2106/RadGPT/MiniGPT4-v2/llama-2-7b-chat-hf"
|
9 |
-
ckpt: "/ibex/project/c2106/RadGPT/MiniGPT-Med-github/miniGPT_Med_.pth"
|
10 |
-
lora_r: 64
|
11 |
-
lora_alpha: 16
|
12 |
-
|
13 |
-
datasets:
|
14 |
-
cc_sbu_align:
|
15 |
-
vis_processor:
|
16 |
-
train:
|
17 |
-
name: "blip2_image_eval"
|
18 |
-
image_size: 448
|
19 |
-
text_processor:
|
20 |
-
train:
|
21 |
-
name: "blip_caption"
|
22 |
-
|
23 |
-
evaluation_datasets:
|
24 |
-
rsna:
|
25 |
-
eval_file_path: miniGPT-Med/json_files/RSNA/full_RSNA_1024.json
|
26 |
-
img_path: miniGPT-Med/RSNA/RSNA-bbox-1024
|
27 |
-
max_new_tokens: 100
|
28 |
-
batch_size: 10
|
29 |
-
|
30 |
-
radvqa:
|
31 |
-
eval_file_path: /miniGPT-Med/json_files/vqa/full_radVQA.json
|
32 |
-
img_path: /miniGPT-Med/radVQA/VQA_RAD_Images
|
33 |
-
max_new_tokens: 300
|
34 |
-
batch_size: 10
|
35 |
-
|
36 |
-
mimic_cxr:
|
37 |
-
eval_file_path: /miniGPT-Med/json_files/mimic/MIMIC_test.json
|
38 |
-
img_path: /miniGPT-Med/mimic-cxr-dataset/image
|
39 |
-
max_new_tokens: 300
|
40 |
-
batch_size: 10
|
41 |
-
|
42 |
-
nlst:
|
43 |
-
eval_file_path: /miniGPT-Med/json_files/NLST/NLST_test.json
|
44 |
-
img_path: /miniGPT-Med/NLST/NLST_images
|
45 |
-
max_new_tokens: 100
|
46 |
-
batch_size: 10
|
47 |
-
|
48 |
-
detect_mimic:
|
49 |
-
eval_file_path: /miniGPT-Med/json_files/MIMIC-bbox/MIMIC-benchmarck.json
|
50 |
-
img_path: /miniGPT-Med/mimic-cxr-dataset/image
|
51 |
-
max_new_tokens: 100
|
52 |
-
batch_size: 10
|
53 |
-
|
54 |
-
SLAKE:
|
55 |
-
eval_file_path: /miniGPT-Med/json_files/SLAKE/grounding_test_SLAKE.json
|
56 |
-
img_path: /miniGPT-Med/SLAKE_images/imgs
|
57 |
-
max_new_tokens: 100
|
58 |
-
batch_size: 10
|
59 |
-
|
60 |
-
|
61 |
-
run:
|
62 |
-
task: image_text_pretrain
|
63 |
-
name: minigptv2_evaluation
|
64 |
-
save_path: /miniGPT-Med/expermints
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/eval_configs/minigptv2_eval.yaml
DELETED
@@ -1,24 +0,0 @@
|
|
1 |
-
model:
|
2 |
-
arch: minigpt_v2
|
3 |
-
model_type: pretrain
|
4 |
-
max_txt_len: 500
|
5 |
-
end_sym: "</s>"
|
6 |
-
low_resource: True
|
7 |
-
prompt_template: '[INST] {} [/INST]'
|
8 |
-
ckpt: "/ibex/project/c2106/RadGPT/MiniGPT-Med-github/miniGPT_Med_.pth"
|
9 |
-
lora_r: 64
|
10 |
-
lora_alpha: 16
|
11 |
-
|
12 |
-
|
13 |
-
datasets:
|
14 |
-
cc_sbu_align:
|
15 |
-
vis_processor:
|
16 |
-
train:
|
17 |
-
name: "blip2_image_eval"
|
18 |
-
image_size: 448
|
19 |
-
text_processor:
|
20 |
-
train:
|
21 |
-
name: "blip_caption"
|
22 |
-
|
23 |
-
run:
|
24 |
-
task: image_text_pretrain
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/eval_scripts/.DS_Store
DELETED
Binary file (6.15 kB)
|
|
MiniGPT-Med-github/eval_scripts/__pycache__/IoU.cpython-39.pyc
DELETED
Binary file (1.77 kB)
|
|
MiniGPT-Med-github/eval_scripts/__pycache__/clean_json.cpython-39.pyc
DELETED
Binary file (2.08 kB)
|
|
MiniGPT-Med-github/eval_scripts/__pycache__/metrics.cpython-39.pyc
DELETED
Binary file (4.28 kB)
|
|
MiniGPT-Med-github/eval_scripts/clean_json.py
DELETED
@@ -1,74 +0,0 @@
|
|
1 |
-
import json
|
2 |
-
import re
|
3 |
-
|
4 |
-
def clean_mimic_json(messy_json, cleaned_output):
|
5 |
-
with open(messy_json, 'r') as f:
|
6 |
-
messy_data = json.load(f)
|
7 |
-
|
8 |
-
clean_data = []
|
9 |
-
for image_id, captions in messy_data.items():
|
10 |
-
image_id_clean = image_id.split('.')[0]
|
11 |
-
caption_clean = ' '.join(captions)
|
12 |
-
|
13 |
-
clean_item = {
|
14 |
-
"image_id": image_id_clean,
|
15 |
-
"caption": caption_clean
|
16 |
-
}
|
17 |
-
|
18 |
-
clean_data.append(clean_item)
|
19 |
-
|
20 |
-
with open(cleaned_output, 'w') as outfile:
|
21 |
-
json.dump(clean_data, outfile, indent=2)
|
22 |
-
|
23 |
-
|
24 |
-
def clean_vqa_json(messy_json, cleaned_output):
|
25 |
-
with open(messy_json, "r") as file:
|
26 |
-
messy_json = json.load(file)
|
27 |
-
|
28 |
-
organized_json = {}
|
29 |
-
|
30 |
-
for key, values in messy_json.items():
|
31 |
-
organized_json[key] = []
|
32 |
-
for value in values:
|
33 |
-
organized_json[key].append({
|
34 |
-
"question": value["question"],
|
35 |
-
"answer": value["answer"]
|
36 |
-
})
|
37 |
-
|
38 |
-
with open(cleaned_output, "w") as outfile:
|
39 |
-
json.dump(organized_json, outfile, indent=4)
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
def clean_detection_json(messy_json, cleaned_output):
|
44 |
-
|
45 |
-
with open(messy_json, "r") as input_file:
|
46 |
-
input_json = json.load(input_file)
|
47 |
-
|
48 |
-
organized_data = []
|
49 |
-
|
50 |
-
for key, value in input_json.items():
|
51 |
-
if value and isinstance(value, list) and len(value) > 0:
|
52 |
-
caption = value[0]
|
53 |
-
objects_match = caption.split("<p>")
|
54 |
-
if len(objects_match) == 2:
|
55 |
-
object_part = objects_match[1].split("</p>")[0].strip()
|
56 |
-
else:
|
57 |
-
object_part = ""
|
58 |
-
|
59 |
-
bbox_match = re.findall(r'<(\d+)>', caption)
|
60 |
-
|
61 |
-
if object_part and bbox_match and len(bbox_match) == 4:
|
62 |
-
key_part = key.split(".png")[0]
|
63 |
-
bbox_values = [float(val) for val in bbox_match]
|
64 |
-
|
65 |
-
organized_item = {
|
66 |
-
"key": key_part,
|
67 |
-
"objects": [object_part],
|
68 |
-
"bbox": [bbox_values],
|
69 |
-
}
|
70 |
-
|
71 |
-
organized_data.append(organized_item)
|
72 |
-
|
73 |
-
with open(cleaned_output, "w") as output_file:
|
74 |
-
json.dump(organized_data, output_file, indent=4)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/eval_scripts/metrics.py
DELETED
@@ -1,164 +0,0 @@
|
|
1 |
-
import sys
|
2 |
-
sys.path.append('.')
|
3 |
-
|
4 |
-
import json
|
5 |
-
import pandas as pd
|
6 |
-
import csv
|
7 |
-
from sentence_transformers import SentenceTransformer, util
|
8 |
-
from minigpt4.common.eval_utils import computeIoU
|
9 |
-
|
10 |
-
# Load pre-trained BERT model
|
11 |
-
model = SentenceTransformer('paraphrase-MiniLM-L6-v2')
|
12 |
-
|
13 |
-
|
14 |
-
# BERT similarity function will be utilized in the two following functions
|
15 |
-
def compute_bert_similarity(prediction_caption, ground_truth_caption):
|
16 |
-
prediction_embedding = model.encode([prediction_caption])
|
17 |
-
ground_truth_embedding = model.encode([ground_truth_caption])
|
18 |
-
similarity = util.pytorch_cos_sim(prediction_embedding, ground_truth_embedding)[0][0].item()
|
19 |
-
return similarity
|
20 |
-
|
21 |
-
|
22 |
-
def MIMIC_BERT_Sim(gt_pth, pred_pth, output_csv):
|
23 |
-
# Read the ground truth and prediction JSON files
|
24 |
-
with open(gt_pth, 'r') as f:
|
25 |
-
ground_truth_data = json.load(f)
|
26 |
-
|
27 |
-
with open(pred_pth, 'r') as f:
|
28 |
-
prediction_data = json.load(f)
|
29 |
-
|
30 |
-
# Create a list to store BERT similarity data
|
31 |
-
bert_similarity_data = []
|
32 |
-
|
33 |
-
# Initialize variables to calculate the average
|
34 |
-
total_similarity = 0
|
35 |
-
total_count = 0
|
36 |
-
|
37 |
-
# Iterate over each item in the prediction_data list
|
38 |
-
for item in prediction_data:
|
39 |
-
# Extract the image_id and corresponding prediction caption
|
40 |
-
image_id = item["image_id"]
|
41 |
-
prediction_caption = item["caption"]
|
42 |
-
|
43 |
-
# Search for the matching ground truth caption based on image_id
|
44 |
-
ground_truth_caption = None
|
45 |
-
for gt_item in ground_truth_data:
|
46 |
-
if gt_item["image_id"] == image_id:
|
47 |
-
ground_truth_caption = gt_item["caption"]
|
48 |
-
break
|
49 |
-
|
50 |
-
if ground_truth_caption is not None:
|
51 |
-
bert_similarity = compute_bert_similarity(prediction_caption, ground_truth_caption)
|
52 |
-
bert_similarity_data.append({"image_id": image_id, "BERT_score": bert_similarity})
|
53 |
-
|
54 |
-
total_similarity += bert_similarity
|
55 |
-
total_count += 1
|
56 |
-
|
57 |
-
average_similarity = total_similarity / total_count if total_count > 0 else 0
|
58 |
-
|
59 |
-
df = pd.DataFrame(bert_similarity_data)
|
60 |
-
df_sorted = df.sort_values(by="BERT_score", ascending=True)
|
61 |
-
df_sorted.to_csv(output_csv, index=False)
|
62 |
-
|
63 |
-
return average_similarity
|
64 |
-
|
65 |
-
def VQA_BERT_Sim(gt_pth, pred_pth, output_csv):
|
66 |
-
# Load ground truth JSON file
|
67 |
-
with open(gt_pth, 'r') as file:
|
68 |
-
gt_data = json.load(file)
|
69 |
-
|
70 |
-
# Load prediction JSON file
|
71 |
-
with open(pred_pth, 'r') as file:
|
72 |
-
prediction_data = json.load(file)
|
73 |
-
|
74 |
-
gt_qa_pairs = {(entry['image_name'], entry['question']): entry['answer'] for entry in gt_data}
|
75 |
-
|
76 |
-
def convert_to_dict(data):
|
77 |
-
qa_dict = {}
|
78 |
-
for image_name, qa_list in data.items():
|
79 |
-
for qa in qa_list:
|
80 |
-
key = (image_name, qa['question'])
|
81 |
-
qa_dict[key] = qa['answer']
|
82 |
-
return qa_dict
|
83 |
-
|
84 |
-
pred_qa_dict = convert_to_dict(prediction_data)
|
85 |
-
|
86 |
-
# Compute BERT similarity and create a list of results
|
87 |
-
results = []
|
88 |
-
|
89 |
-
for key, gt_answer in gt_qa_pairs.items():
|
90 |
-
if key in pred_qa_dict:
|
91 |
-
pred_answer = pred_qa_dict[key]
|
92 |
-
gt_answer = str(gt_answer)
|
93 |
-
pred_answer = str(pred_answer)
|
94 |
-
|
95 |
-
# Compute BERT similarity
|
96 |
-
similarity_score = compute_bert_similarity(pred_answer, gt_answer)
|
97 |
-
|
98 |
-
# Append the result to the list
|
99 |
-
results.append({
|
100 |
-
"img_name": key[0],
|
101 |
-
"question": key[1],
|
102 |
-
"answer": pred_answer,
|
103 |
-
"BERT_score": similarity_score
|
104 |
-
})
|
105 |
-
|
106 |
-
average_similarity = sum(entry["BERT_score"] for entry in results) / len(results) if results else 0
|
107 |
-
df = pd.DataFrame(results)
|
108 |
-
df_sorted = df.sort_values(by="BERT_score", ascending=True)
|
109 |
-
df_sorted.to_csv(output_csv, index=False)
|
110 |
-
print(f"Average BERT similarity score: {average_similarity}")
|
111 |
-
|
112 |
-
|
113 |
-
#################################
|
114 |
-
##############IoU################
|
115 |
-
#################################
|
116 |
-
|
117 |
-
def preprocess_bbox(bbox, original_size, image_size):
|
118 |
-
x1 = int((bbox[0] / original_size) * image_size)
|
119 |
-
y1 = int((bbox[1] / original_size) * image_size)
|
120 |
-
x2 = int((bbox[2] / original_size) * image_size)
|
121 |
-
y2 = int((bbox[3] / original_size) * image_size)
|
122 |
-
return [x1, y1, x2, y2]
|
123 |
-
|
124 |
-
def average_iou(gt_pth, pred_pth, original_size, image_size, dataset_name, csv_filename):
|
125 |
-
# Load ground truth
|
126 |
-
with open(gt_pth, 'r') as file:
|
127 |
-
ground_truth = json.load(file)
|
128 |
-
|
129 |
-
# Load predictions
|
130 |
-
with open(pred_pth, 'r') as file:
|
131 |
-
predictions = json.load(file)
|
132 |
-
|
133 |
-
iou_list = []
|
134 |
-
|
135 |
-
with open(csv_filename, 'w', newline='') as csvfile:
|
136 |
-
fieldnames = ['image_name', 'IoU']
|
137 |
-
writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
|
138 |
-
writer.writeheader()
|
139 |
-
|
140 |
-
for gt_item in ground_truth:
|
141 |
-
gt_key = gt_item['key']
|
142 |
-
gt_bboxes = gt_item['bbox']
|
143 |
-
original_size = gt_item['height']
|
144 |
-
gt_processed_bboxes = [preprocess_bbox(bbox, original_size, image_size) for bbox in gt_bboxes]
|
145 |
-
|
146 |
-
for pred_item in predictions:
|
147 |
-
pred_key = pred_item['key'].replace(".png", "")
|
148 |
-
|
149 |
-
if gt_key == pred_key:
|
150 |
-
pred_bboxes = pred_item['bbox']
|
151 |
-
try:
|
152 |
-
for gt_bbox in gt_processed_bboxes:
|
153 |
-
for pred_bbox in pred_bboxes:
|
154 |
-
iou = computeIoU(gt_bbox, pred_bbox)
|
155 |
-
iou_list.append(iou)
|
156 |
-
writer.writerow({'image_name': gt_key, 'IoU': iou})
|
157 |
-
print(gt_key)
|
158 |
-
print(iou)
|
159 |
-
except Exception as e:
|
160 |
-
print("gt_bbox: ", gt_bbox)
|
161 |
-
print("gt_bbox: ", pred_bboxes)
|
162 |
-
|
163 |
-
# average_iou = sum(iou_list) / len(iou_list)
|
164 |
-
# print(f"Average IoU for dataset {dataset_name}: {average_iou:.4f}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/eval_scripts/model_evaluation.py
DELETED
@@ -1,274 +0,0 @@
|
|
1 |
-
'''
|
2 |
-
use this command in terminal to run the evaluation script
|
3 |
-
torchrun --master-port 8888 --nproc_per_node 1 eval_scripts/model_evaluation.py --cfg-path eval_configs/minigptv2_benchmark_evaluation.yaml --dataset
|
4 |
-
|
5 |
-
|
6 |
-
'''
|
7 |
-
|
8 |
-
import sys
|
9 |
-
sys.path.append('.')
|
10 |
-
import os
|
11 |
-
import re
|
12 |
-
import json
|
13 |
-
import argparse
|
14 |
-
from collections import defaultdict
|
15 |
-
import random
|
16 |
-
import numpy as np
|
17 |
-
from PIL import Image
|
18 |
-
from tqdm import tqdm
|
19 |
-
import torch
|
20 |
-
from torch.utils.data import DataLoader
|
21 |
-
from minigpt4.common.config import Config
|
22 |
-
from minigpt4.common.eval_utils import prepare_texts, init_model, eval_parser, computeIoU
|
23 |
-
from minigpt4.conversation.conversation import CONV_VISION_minigptv2
|
24 |
-
|
25 |
-
from minigpt4.datasets.datasets.mimic_cxr_dataset import evalMIMICDataset, evalDetectMimicDataset
|
26 |
-
from minigpt4.datasets.datasets.radvqa_dataset import evalRadVQADataset
|
27 |
-
from minigpt4.datasets.datasets.nlst_dataset import eval_NLST_Dataset
|
28 |
-
from minigpt4.datasets.datasets.rsna_dataset import evalRSNADataset
|
29 |
-
from minigpt4.datasets.datasets.SLAKE_dataset import evalSLAKEDataset
|
30 |
-
#import cleaning classes
|
31 |
-
from eval_scripts.clean_json import clean_mimic_json, clean_vqa_json, clean_detection_json
|
32 |
-
from eval_scripts.metrics import MIMIC_BERT_Sim, VQA_BERT_Sim, average_iou
|
33 |
-
|
34 |
-
def list_of_str(arg):
|
35 |
-
return list(map(str, arg.split(',')))
|
36 |
-
|
37 |
-
parser = eval_parser()
|
38 |
-
parser.add_argument("--dataset", type=list_of_str, help="dataset to evaluate")
|
39 |
-
|
40 |
-
args = parser.parse_args()
|
41 |
-
|
42 |
-
cfg = Config(args)
|
43 |
-
|
44 |
-
|
45 |
-
model, vis_processor = init_model(args)
|
46 |
-
model.eval()
|
47 |
-
CONV_VISION = CONV_VISION_minigptv2
|
48 |
-
conv_temp = CONV_VISION.copy()
|
49 |
-
conv_temp.system = ""
|
50 |
-
model.eval()
|
51 |
-
save_path = cfg.run_cfg.save_path
|
52 |
-
|
53 |
-
def process_mimic_dataset():
|
54 |
-
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
|
55 |
-
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
|
56 |
-
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
|
57 |
-
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
|
58 |
-
|
59 |
-
with open((eval_file_path), 'r') as f:
|
60 |
-
mimic = json.load(f)
|
61 |
-
|
62 |
-
data = evalMIMICDataset(mimic, vis_processor, img_path)
|
63 |
-
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
|
64 |
-
minigpt4_predict = defaultdict(list)
|
65 |
-
|
66 |
-
for images, questions, img_ids in tqdm(eval_dataloader):
|
67 |
-
texts = prepare_texts(questions, conv_temp) # warp the texts with conversation template
|
68 |
-
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
|
69 |
-
for answer, img_id, question in zip(answers, img_ids, questions):
|
70 |
-
minigpt4_predict[img_id].append(answer)
|
71 |
-
|
72 |
-
file_save_path = os.path.join(save_path,"MIMIC_inference_results_stage3.json")
|
73 |
-
with open(file_save_path,'w') as f:
|
74 |
-
json.dump(minigpt4_predict, f)
|
75 |
-
clean_mimic_json(file_save_path, file_save_path)
|
76 |
-
|
77 |
-
# csv file path to save the BERT results per each case
|
78 |
-
output_csv_path = '/miniGPT-Med/metric_results/bert_similarity_scores.csv'
|
79 |
-
|
80 |
-
# in MIMIC_BERT_Sim add the path of the ground_truth then the path of the inference result
|
81 |
-
average_similarity = MIMIC_BERT_Sim(eval_file_path, file_save_path, output_csv_path)
|
82 |
-
#print the average BERT_Sim
|
83 |
-
print("Average BERT Similarity:", average_similarity)
|
84 |
-
|
85 |
-
def process_vqa_dataset():
|
86 |
-
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
|
87 |
-
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
|
88 |
-
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
|
89 |
-
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
|
90 |
-
|
91 |
-
with open((eval_file_path), 'r') as f:
|
92 |
-
radVQA = json.load(f)
|
93 |
-
|
94 |
-
data = evalRadVQADataset(radVQA, vis_processor, img_path)
|
95 |
-
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
|
96 |
-
minigpt4_predict = defaultdict(list)
|
97 |
-
|
98 |
-
for images, questions, img_ids in tqdm(eval_dataloader):
|
99 |
-
texts = prepare_texts(questions, conv_temp) # warp the texts with conversation template
|
100 |
-
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
|
101 |
-
for answer, img_id, question in zip(answers, img_ids, questions):
|
102 |
-
minigpt4_predict[img_id].append({"key":img_ids,"question": question.replace("[vqa]", "").strip() , "answer": answer})
|
103 |
-
|
104 |
-
file_save_path = os.path.join(save_path,"radVQA_inference_results.json")
|
105 |
-
output_csv_path = '/miniGPT-Med/BERT_Sim_results/vqa_bert_similarity_scores.csv'
|
106 |
-
|
107 |
-
with open(file_save_path,'w') as f:
|
108 |
-
json.dump(minigpt4_predict, f)
|
109 |
-
|
110 |
-
clean_vqa_json(file_save_path, file_save_path)
|
111 |
-
VQA_BERT_Sim(eval_file_path, file_save_path, output_csv_path)
|
112 |
-
|
113 |
-
def process_nlst_dataset():
|
114 |
-
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
|
115 |
-
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
|
116 |
-
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
|
117 |
-
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
|
118 |
-
|
119 |
-
with open((eval_file_path), 'r') as f:
|
120 |
-
nlst = json.load(f)
|
121 |
-
|
122 |
-
data = eval_NLST_Dataset(nlst, vis_processor, img_path)
|
123 |
-
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
|
124 |
-
minigpt4_predict = defaultdict(list)
|
125 |
-
resamples = []
|
126 |
-
|
127 |
-
for images, questions, img_ids in tqdm(eval_dataloader):
|
128 |
-
|
129 |
-
texts = prepare_texts(questions, conv_temp)
|
130 |
-
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
|
131 |
-
|
132 |
-
for answer, img_id, question in zip(answers, img_ids, questions):
|
133 |
-
|
134 |
-
# answer = answer.replace("<unk>","").replace(" ","").strip()
|
135 |
-
pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
|
136 |
-
minigpt4_predict[img_id].append(answer)
|
137 |
-
|
138 |
-
file_save_path = os.path.join(save_path,"NLST_inference_result.json")
|
139 |
-
with open(file_save_path,'w') as f:
|
140 |
-
json.dump(minigpt4_predict, f)
|
141 |
-
|
142 |
-
csv_pth = os.path.join(save_path,"NLST_IoU_results.csv")
|
143 |
-
clean_detection_json(file_save_path,file_save_path)
|
144 |
-
average_iou(eval_file_path, file_save_path, 512, 100, "NLST", csv_pth)
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
def process_rsna_dataset():
|
149 |
-
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
|
150 |
-
print(eval_file_path)
|
151 |
-
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
|
152 |
-
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
|
153 |
-
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
|
154 |
-
print("----config----")
|
155 |
-
with open((eval_file_path), 'r') as f:
|
156 |
-
nlst = json.load(f)
|
157 |
-
|
158 |
-
data = evalRSNADataset(nlst, vis_processor, img_path)
|
159 |
-
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
|
160 |
-
minigpt4_predict = defaultdict(list)
|
161 |
-
resamples = []
|
162 |
-
|
163 |
-
for images, questions, img_ids in tqdm(eval_dataloader):
|
164 |
-
texts = prepare_texts(questions, conv_temp)
|
165 |
-
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
|
166 |
-
|
167 |
-
for answer, img_id, question in zip(answers, img_ids, questions):
|
168 |
-
|
169 |
-
# answer = answer.replace("<unk>","").replace(" ","").strip()
|
170 |
-
pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
|
171 |
-
minigpt4_predict[img_id].append(answer)
|
172 |
-
print(img_id)
|
173 |
-
print(answer)
|
174 |
-
|
175 |
-
file_save_path = os.path.join(save_path,"RSNA_inference_result.json")
|
176 |
-
with open(file_save_path,'w') as f:
|
177 |
-
json.dump(minigpt4_predict, f)
|
178 |
-
|
179 |
-
csv_pth = os.path.join(save_path,"RSNA_IoU_results.csv")
|
180 |
-
clean_detection_json(file_save_path,file_save_path)
|
181 |
-
average_iou(eval_file_path, file_save_path, 1024, 100, "rsna", csv_pth)
|
182 |
-
|
183 |
-
|
184 |
-
def process_detect_mimic():
|
185 |
-
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
|
186 |
-
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
|
187 |
-
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
|
188 |
-
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
|
189 |
-
|
190 |
-
with open((eval_file_path), 'r') as f:
|
191 |
-
nlst = json.load(f)
|
192 |
-
|
193 |
-
data = evalDetectMimicDataset(nlst, vis_processor, img_path)
|
194 |
-
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
|
195 |
-
minigpt4_predict = defaultdict(list)
|
196 |
-
resamples = []
|
197 |
-
|
198 |
-
for images, questions, img_ids in tqdm(eval_dataloader):
|
199 |
-
|
200 |
-
texts = prepare_texts(questions, conv_temp)
|
201 |
-
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
|
202 |
-
|
203 |
-
for answer, img_id, question in zip(answers, img_ids, questions):
|
204 |
-
pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
|
205 |
-
minigpt4_predict[img_id].append(answer)
|
206 |
-
|
207 |
-
file_save_path = os.path.join(save_path,"Detect_MIMIC_inference_result.json")
|
208 |
-
with open(file_save_path,'w') as f:
|
209 |
-
json.dump(minigpt4_predict, f)
|
210 |
-
|
211 |
-
|
212 |
-
csv_pth = os.path.join(save_path,"MIMIC_IoU_results.csv")
|
213 |
-
clean_detection_json(file_save_path,file_save_path)
|
214 |
-
average_iou(eval_file_path, file_save_path, "to be specified soon", 100, "MIMIC", csv_pth)
|
215 |
-
|
216 |
-
|
217 |
-
|
218 |
-
def process_SLAKE_dataset():
|
219 |
-
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
|
220 |
-
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
|
221 |
-
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
|
222 |
-
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
|
223 |
-
|
224 |
-
with open((eval_file_path), 'r') as f:
|
225 |
-
SLAKE = json.load(f)
|
226 |
-
|
227 |
-
data = evalSLAKEDataset(SLAKE, vis_processor, img_path)
|
228 |
-
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
|
229 |
-
minigpt4_predict = defaultdict(list)
|
230 |
-
resamples = []
|
231 |
-
|
232 |
-
for images, questions, img_ids in tqdm(eval_dataloader):
|
233 |
-
|
234 |
-
texts = prepare_texts(questions, conv_temp)
|
235 |
-
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
|
236 |
-
|
237 |
-
for answer, img_id, question in zip(answers, img_ids, questions):
|
238 |
-
|
239 |
-
# answer = answer.replace("<unk>","").replace(" ","").strip()
|
240 |
-
pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
|
241 |
-
minigpt4_predict[img_id].append(answer)
|
242 |
-
|
243 |
-
file_save_path = os.path.join(save_path,"SLAKE_inference_result.json")
|
244 |
-
with open(file_save_path,'w') as f:
|
245 |
-
json.dump(minigpt4_predict, f)
|
246 |
-
|
247 |
-
csv_pth = os.path.join(save_path,"SLAKE_IoU_results.csv")
|
248 |
-
clean_detection_json(file_save_path,file_save_path)
|
249 |
-
average_iou(eval_file_path, file_save_path, 100, 100, "SLAKE", csv_pth)
|
250 |
-
|
251 |
-
|
252 |
-
|
253 |
-
############################################################################
|
254 |
-
for dataset in args.dataset:
|
255 |
-
if dataset == 'mimic_cxr':
|
256 |
-
process_mimic_dataset()
|
257 |
-
|
258 |
-
elif dataset == 'radvqa':
|
259 |
-
process_vqa_dataset()
|
260 |
-
|
261 |
-
elif dataset == 'nlst':
|
262 |
-
process_nlst_dataset()
|
263 |
-
|
264 |
-
elif dataset == 'rsna':
|
265 |
-
process_rsna_dataset()
|
266 |
-
|
267 |
-
elif dataset == 'detect_mimic':
|
268 |
-
process_detect_mimic()
|
269 |
-
|
270 |
-
elif dataset == 'SLAKE':
|
271 |
-
process_SLAKE_dataset()
|
272 |
-
|
273 |
-
else:
|
274 |
-
print(f"Dataset '{dataset}' is not supported.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/miniGPTV2.yml
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
name: GPTv2
|
2 |
-
channels:
|
3 |
-
- pytorch
|
4 |
-
- defaults
|
5 |
-
- anaconda
|
6 |
-
dependencies:
|
7 |
-
- python=3.9
|
8 |
-
- cudatoolkit
|
9 |
-
- pip
|
10 |
-
- pip:
|
11 |
-
- torch==2.0.0
|
12 |
-
- torchaudio
|
13 |
-
- torchvision
|
14 |
-
- huggingface-hub==0.18.0
|
15 |
-
- matplotlib==3.7.0
|
16 |
-
- psutil==5.9.4
|
17 |
-
- iopath
|
18 |
-
- pyyaml==6.0
|
19 |
-
- regex==2022.10.31
|
20 |
-
- tokenizers==0.13.2
|
21 |
-
- tqdm==4.64.1
|
22 |
-
- transformers==4.30.0
|
23 |
-
- timm==0.6.13
|
24 |
-
- webdataset==0.2.48
|
25 |
-
- omegaconf==2.3.0
|
26 |
-
- opencv-python==4.7.0.72
|
27 |
-
- decord==0.6.0
|
28 |
-
- peft==0.2.0
|
29 |
-
- sentence-transformers
|
30 |
-
- gradio==3.47.1
|
31 |
-
- accelerate==0.20.3
|
32 |
-
- bitsandbytes==0.37.0
|
33 |
-
- scikit-image
|
34 |
-
- visual-genome
|
35 |
-
- wandb
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/miniGPT_Med_.pth
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:ca2d7fc37dc5330cdae927c8a3ff649c5919c726eccb05cae921fb997028b08e
|
3 |
-
size 679780138
|
|
|
|
|
|
|
|
MiniGPT-Med-github/minigpt4/.DS_Store
DELETED
Binary file (6.15 kB)
|
|
MiniGPT-Med-github/minigpt4/__init__.py
DELETED
@@ -1,31 +0,0 @@
|
|
1 |
-
"""
|
2 |
-
Copyright (c) 2022, salesforce.com, inc.
|
3 |
-
All rights reserved.
|
4 |
-
SPDX-License-Identifier: BSD-3-Clause
|
5 |
-
For full license text, see the LICENSE_Lavis file in the repo root or https://opensource.org/licenses/BSD-3-Clause
|
6 |
-
"""
|
7 |
-
|
8 |
-
import os
|
9 |
-
import sys
|
10 |
-
|
11 |
-
from omegaconf import OmegaConf
|
12 |
-
|
13 |
-
from minigpt4.common.registry import registry
|
14 |
-
|
15 |
-
from minigpt4.datasets.builders import *
|
16 |
-
from minigpt4.models import *
|
17 |
-
from minigpt4.processors import *
|
18 |
-
from minigpt4.tasks import *
|
19 |
-
|
20 |
-
|
21 |
-
root_dir = os.path.dirname(os.path.abspath(__file__))
|
22 |
-
default_cfg = OmegaConf.load(os.path.join(root_dir, "configs/default.yaml"))
|
23 |
-
|
24 |
-
registry.register_path("library_root", root_dir)
|
25 |
-
repo_root = os.path.join(root_dir, "..")
|
26 |
-
registry.register_path("repo_root", repo_root)
|
27 |
-
cache_root = os.path.join(repo_root, default_cfg.env.cache_root)
|
28 |
-
registry.register_path("cache_root", cache_root)
|
29 |
-
|
30 |
-
registry.register("MAX_INT", sys.maxsize)
|
31 |
-
registry.register("SPLIT_NAMES", ["train", "val", "test"])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MiniGPT-Med-github/minigpt4/__pycache__/__init__.cpython-310.pyc
DELETED
Binary file (1.03 kB)
|
|
MiniGPT-Med-github/minigpt4/__pycache__/__init__.cpython-39.pyc
DELETED
Binary file (1.03 kB)
|
|
MiniGPT-Med-github/minigpt4/common/.DS_Store
DELETED
Binary file (6.15 kB)
|
|
MiniGPT-Med-github/minigpt4/common/__init__.py
DELETED
File without changes
|
MiniGPT-Med-github/minigpt4/common/__pycache__/__init__.cpython-310.pyc
DELETED
Binary file (166 Bytes)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/__init__.cpython-39.pyc
DELETED
Binary file (164 Bytes)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/config.cpython-310.pyc
DELETED
Binary file (12.6 kB)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/config.cpython-39.pyc
DELETED
Binary file (12.7 kB)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/dist_utils.cpython-310.pyc
DELETED
Binary file (3.81 kB)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/dist_utils.cpython-39.pyc
DELETED
Binary file (3.82 kB)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/eval_utils.cpython-39.pyc
DELETED
Binary file (3.17 kB)
|
|
MiniGPT-Med-github/minigpt4/common/__pycache__/logger.cpython-310.pyc
DELETED
Binary file (6.43 kB)
|
|