README.md

---
library_name: pytorch
license: mit
tags:
- android
pipeline_tag: image-to-text

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/trocr/web-assets/model_demo.png)

# TrOCR: Optimized for Mobile Deployment
## Transformer based model for state-of-the-art optical character recognition (OCR) on both printed and handwritten text


End-to-end text recognition approach with pre-trained image transformer and text transformer models for both image understanding and wordpiece-level text generation.

This model is an implementation of TrOCR found [here](https://huggingface.co/microsoft/trocr-small-stage1).


This repository provides scripts to run TrOCR on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/trocr).


### Model Details

- **Model Type:** Image to text
- **Model Stats:**
  - Model checkpoint: trocr-small-stage1
  - Input resolution: 320x320
  - Number of parameters (TrOCREncoder): 23.0M
  - Model size (TrOCREncoder): 87.8 MB
  - Number of parameters (TrOCRDecoder): 38.3M
  - Model size (TrOCRDecoder): 146 MB

| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| TrOCRDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 2.227 ms | 0 - 353 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 2.276 ms | 2 - 4 MB | FP16 | NPU | [TrOCR.so](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.so) |
| TrOCRDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 2.926 ms | 0 - 170 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.onnx) |
| TrOCRDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 1.587 ms | 0 - 54 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 1.665 ms | 0 - 15 MB | FP16 | NPU | [TrOCR.so](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.so) |
| TrOCRDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 2.118 ms | 0 - 64 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.onnx) |
| TrOCRDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 1.489 ms | 0 - 48 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.595 ms | 3 - 51 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 1.87 ms | 2 - 47 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.onnx) |
| TrOCRDecoder | SA7255P ADP | SA7255P | TFLITE | 12.182 ms | 0 - 43 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | SA7255P ADP | SA7255P | QNN | 12.308 ms | 7 - 14 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 2.193 ms | 0 - 289 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 2.487 ms | 2 - 5 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | SA8295P ADP | SA8295P | TFLITE | 3.154 ms | 0 - 41 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | SA8295P ADP | SA8295P | QNN | 3.561 ms | 7 - 17 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 2.185 ms | 0 - 308 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 2.255 ms | 2 - 4 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | SA8775P ADP | SA8775P | TFLITE | 3.366 ms | 0 - 43 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | SA8775P ADP | SA8775P | QNN | 3.497 ms | 7 - 14 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 12.182 ms | 0 - 43 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | QCS8275 (Proxy) | QCS8275 Proxy | QNN | 12.308 ms | 7 - 14 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 2.189 ms | 0 - 87 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 2.308 ms | 2 - 5 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 3.366 ms | 0 - 43 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | QCS9075 (Proxy) | QCS9075 Proxy | QNN | 3.497 ms | 7 - 14 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 2.479 ms | 0 - 49 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.tflite) |
| TrOCRDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 2.743 ms | 5 - 54 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 2.406 ms | 7 - 7 MB | FP16 | NPU | Use Export Script |
| TrOCRDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2.663 ms | 68 - 68 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCRDecoder.onnx) |
| TrOCREncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 50.264 ms | 7 - 31 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 37.193 ms | 2 - 4 MB | FP16 | NPU | [TrOCR.so](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.so) |
| TrOCREncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 38.277 ms | 14 - 161 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.onnx) |
| TrOCREncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 39.543 ms | 4 - 65 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 29.756 ms | 2 - 18 MB | FP16 | NPU | [TrOCR.so](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.so) |
| TrOCREncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 32.2 ms | 16 - 77 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.onnx) |
| TrOCREncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 32.147 ms | 6 - 70 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 37.162 ms | 2 - 66 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 27.594 ms | 8 - 70 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.onnx) |
| TrOCREncoder | SA7255P ADP | SA7255P | TFLITE | 266.206 ms | 1 - 63 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | SA7255P ADP | SA7255P | QNN | 247.7 ms | 2 - 9 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 50.096 ms | 7 - 31 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 37.306 ms | 2 - 4 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | SA8295P ADP | SA8295P | TFLITE | 65.436 ms | 7 - 67 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | SA8295P ADP | SA8295P | QNN | 50.625 ms | 2 - 12 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 50.166 ms | 7 - 27 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 37.292 ms | 2 - 4 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | SA8775P ADP | SA8775P | TFLITE | 59.798 ms | 7 - 69 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | SA8775P ADP | SA8775P | QNN | 42.357 ms | 2 - 10 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 266.206 ms | 1 - 63 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | QCS8275 (Proxy) | QCS8275 Proxy | QNN | 247.7 ms | 2 - 9 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 50.181 ms | 7 - 35 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 37.081 ms | 2 - 4 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 59.798 ms | 7 - 69 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | QCS9075 (Proxy) | QCS9075 Proxy | QNN | 42.357 ms | 2 - 10 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 60.287 ms | 7 - 66 MB | FP16 | NPU | [TrOCR.tflite](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.tflite) |
| TrOCREncoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 66.83 ms | 2 - 64 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 33.968 ms | 2 - 2 MB | FP16 | NPU | Use Export Script |
| TrOCREncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 36.85 ms | 50 - 50 MB | FP16 | NPU | [TrOCR.onnx](https://huggingface.co/qualcomm/TrOCR/blob/main/TrOCREncoder.onnx) |




## Installation


Install the package via pip:
```bash
pip install "qai-hub-models[trocr]"
```


## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.

With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.



## Demo off target

The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.

```bash
python -m qai_hub_models.models.trocr.demo
```

The above demo runs a reference implementation of pre-processing, model
inference, and post processing.

**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.trocr.demo
```


### Run model on a cloud-hosted device

In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.

```bash
python -m qai_hub_models.models.trocr.export
```
```
Profiling Results
------------------------------------------------------------
TrOCRDecoder
Device                          : Samsung Galaxy S23 (13)
Runtime                         : TFLITE                 
Estimated inference time (ms)   : 2.2                    
Estimated peak memory usage (MB): [0, 353]               
Total # Ops                     : 399                    
Compute Unit(s)                 : NPU (399 ops)          

------------------------------------------------------------
TrOCREncoder
Device                          : Samsung Galaxy S23 (13)
Runtime                         : TFLITE                 
Estimated inference time (ms)   : 50.3                   
Estimated peak memory usage (MB): [7, 31]                
Total # Ops                     : 591                    
Compute Unit(s)                 : NPU (591 ops)          
```


## How does this work?

This [export script](https://aihub.qualcomm.com/models/trocr/qai_hub_models/models/TrOCR/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:

Step 1: **Compile model for on-device deployment**

To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.

```python
import torch

import qai_hub as hub
from qai_hub_models.models.trocr import Model

# Load the model
model = Model.from_pretrained()
decoder_model = model.decoder
encoder_model = model.encoder

# Device
device = hub.Device("Samsung Galaxy S23")

# Trace model
decoder_input_shape = decoder_model.get_input_spec()
decoder_sample_inputs = decoder_model.sample_inputs()

traced_decoder_model = torch.jit.trace(decoder_model, [torch.tensor(data[0]) for _, data in decoder_sample_inputs.items()])

# Compile model on a specific device
decoder_compile_job = hub.submit_compile_job(
    model=traced_decoder_model ,
    device=device,
    input_specs=decoder_model.get_input_spec(),
)

# Get target model to run on-device
decoder_target_model = decoder_compile_job.get_target_model()
# Trace model
encoder_input_shape = encoder_model.get_input_spec()
encoder_sample_inputs = encoder_model.sample_inputs()

traced_encoder_model = torch.jit.trace(encoder_model, [torch.tensor(data[0]) for _, data in encoder_sample_inputs.items()])

# Compile model on a specific device
encoder_compile_job = hub.submit_compile_job(
    model=traced_encoder_model ,
    device=device,
    input_specs=encoder_model.get_input_spec(),
)

# Get target model to run on-device
encoder_target_model = encoder_compile_job.get_target_model()

```


Step 2: **Performance profiling on cloud-hosted device**

After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud.  Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
decoder_profile_job = hub.submit_profile_job(
    model=decoder_target_model,
    device=device,
)
encoder_profile_job = hub.submit_profile_job(
    model=encoder_target_model,
    device=device,
)

```

Step 3: **Verify on-device accuracy**

To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
decoder_input_data = decoder_model.sample_inputs()
decoder_inference_job = hub.submit_inference_job(
    model=decoder_target_model,
    device=device,
    inputs=decoder_input_data,
)
decoder_inference_job.download_output_data()
encoder_input_data = encoder_model.sample_inputs()
encoder_inference_job = hub.submit_inference_job(
    model=encoder_target_model,
    device=device,
    inputs=encoder_input_data,
)
encoder_inference_job.download_output_data()

```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.

**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).




## Deploying compiled model to Android


The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
  tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
  guide to deploy the .tflite model in an Android application.


- QNN (`.so` export ): This [sample
  app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library  in an Android application.


## View on Qualcomm® AI Hub
Get more details on TrOCR's performance across various devices [here](https://aihub.qualcomm.com/models/trocr).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)


## License
* The license for the original implementation of TrOCR can be found
  [here](https://github.com/microsoft/unilm/blob/master/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)



## References
* [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282)
* [Source Model Implementation](https://huggingface.co/microsoft/trocr-small-stage1)



## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:[email protected]).