Commit
·
50cb395
1
Parent(s):
4cf03a8
Updated with OCR model and Gradio integration
Browse files- app.py +16 -13
- requirements.txt +2 -1
app.py
CHANGED
@@ -1,20 +1,27 @@
|
|
1 |
import gradio as gr
|
2 |
-
from
|
3 |
-
from
|
4 |
|
5 |
-
#
|
|
|
|
|
|
|
|
|
6 |
model_name = "OpenGVLab/InternVL2-1B"
|
7 |
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
|
8 |
-
model = AutoModelForVision2Seq.from_pretrained(model_name, trust_remote_code=True)
|
9 |
|
10 |
-
#
|
11 |
-
|
12 |
-
|
|
|
|
|
|
|
|
|
13 |
|
14 |
# Function to process and describe the image
|
15 |
def analyze_image(image):
|
16 |
# Use PIL to load the image
|
17 |
-
img =
|
18 |
# Tokenize the input
|
19 |
inputs = tokenizer("describe this image", return_tensors="pt")
|
20 |
# Perform inference
|
@@ -27,8 +34,4 @@ demo = gr.Interface(
|
|
27 |
inputs=gr.Image(type="pil"), # Upload an image
|
28 |
outputs="text", # Output the extracted text
|
29 |
title="Image Description using OpenGVLab/InternVL2-1B",
|
30 |
-
description="Upload
|
31 |
-
)
|
32 |
-
|
33 |
-
if __name__ == "__main__":
|
34 |
-
demo.launch(share=True)
|
|
|
1 |
import gradio as gr
|
2 |
+
from transformers import AutoTokenizer
|
3 |
+
from huggingface_hub import hf_hub_download
|
4 |
|
5 |
+
# Import the custom model code dynamically
|
6 |
+
import sys
|
7 |
+
sys.path.append(hf_hub_download(repo_id="OpenGVLab/InternVL2-1B", filename="")) # Adjust path
|
8 |
+
|
9 |
+
# Load the custom model and tokenizer
|
10 |
model_name = "OpenGVLab/InternVL2-1B"
|
11 |
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
|
|
|
12 |
|
13 |
+
# Import the custom model class from the downloaded files
|
14 |
+
from transformers_modules.OpenGVLab.InternVL2-1B.configuration_internvl_chat import InternVLChatConfig
|
15 |
+
from transformers_modules.OpenGVLab.InternVL2-1B.modeling_internvl import InternVLForVision2Seq
|
16 |
+
|
17 |
+
# Load the model
|
18 |
+
config = InternVLChatConfig.from_pretrained(model_name, trust_remote_code=True)
|
19 |
+
model = InternVLForVision2Seq.from_pretrained(model_name, config=config, trust_remote_code=True)
|
20 |
|
21 |
# Function to process and describe the image
|
22 |
def analyze_image(image):
|
23 |
# Use PIL to load the image
|
24 |
+
img = image.convert("RGB")
|
25 |
# Tokenize the input
|
26 |
inputs = tokenizer("describe this image", return_tensors="pt")
|
27 |
# Perform inference
|
|
|
34 |
inputs=gr.Image(type="pil"), # Upload an image
|
35 |
outputs="text", # Output the extracted text
|
36 |
title="Image Description using OpenGVLab/InternVL2-1B",
|
37 |
+
description="Upload⬤
|
|
|
|
|
|
|
|
requirements.txt
CHANGED
@@ -5,4 +5,5 @@ gradio
|
|
5 |
datasets
|
6 |
pytesseract
|
7 |
Pillow
|
8 |
-
lmdeploy
|
|
|
|
5 |
datasets
|
6 |
pytesseract
|
7 |
Pillow
|
8 |
+
lmdeploy
|
9 |
+
huggingface_hub
|