Hugging Face's logo

qualcomm
/
TrOCR

Image-to-Text
PyTorch
android
Model card Files
xet
 Community
3
TrOCR / README.md
qaihm-bot's picture
qaihm-bot
Upload README.md with huggingface_hub
c6cde83 verified
preview code
|
raw
history blame
17 kB
metadata 
library_name: pytorch
license: other
tags:
  - android
pipeline_tag: image-to-text

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.

This repository provides scripts to run TrOCR on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

  • Model Type: Model_use_case.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 Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
TrOCRDecoder float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 4.288 ms 0 - 59 MB NPU TrOCR.tflite
TrOCRDecoder float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN 4.124 ms 0 - 9 MB NPU Use Export Script
TrOCRDecoder float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 2.836 ms 0 - 123 MB NPU TrOCR.tflite
TrOCRDecoder float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN 3.059 ms 3 - 124 MB NPU Use Export Script
TrOCRDecoder float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 2.073 ms 0 - 55 MB NPU TrOCR.tflite
TrOCRDecoder float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN 2.031 ms 2 - 5 MB NPU Use Export Script
TrOCRDecoder float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 2.892 ms 0 - 59 MB NPU TrOCR.tflite
TrOCRDecoder float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN 2.752 ms 7 - 17 MB NPU Use Export Script
TrOCRDecoder float SA7255P ADP Qualcomm® SA7255P TFLITE 4.288 ms 0 - 59 MB NPU TrOCR.tflite
TrOCRDecoder float SA7255P ADP Qualcomm® SA7255P QNN 4.124 ms 0 - 9 MB NPU Use Export Script
TrOCRDecoder float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 2.072 ms 0 - 96 MB NPU TrOCR.tflite
TrOCRDecoder float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN 2.079 ms 2 - 4 MB NPU Use Export Script
TrOCRDecoder float SA8295P ADP Qualcomm® SA8295P TFLITE 3.068 ms 0 - 52 MB NPU TrOCR.tflite
TrOCRDecoder float SA8295P ADP Qualcomm® SA8295P QNN 2.882 ms 7 - 25 MB NPU Use Export Script
TrOCRDecoder float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 2.197 ms 0 - 30 MB NPU TrOCR.tflite
TrOCRDecoder float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN 2.145 ms 2 - 4 MB NPU Use Export Script
TrOCRDecoder float SA8775P ADP Qualcomm® SA8775P TFLITE 2.892 ms 0 - 59 MB NPU TrOCR.tflite
TrOCRDecoder float SA8775P ADP Qualcomm® SA8775P QNN 2.752 ms 7 - 17 MB NPU Use Export Script
TrOCRDecoder float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 2.23 ms 0 - 30 MB NPU TrOCR.tflite
TrOCRDecoder float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN 2.038 ms 3 - 26 MB NPU Use Export Script
TrOCRDecoder float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 2.71 ms 0 - 163 MB NPU TrOCR.onnx
TrOCRDecoder float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 1.486 ms 0 - 134 MB NPU TrOCR.tflite
TrOCRDecoder float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN 1.49 ms 0 - 130 MB NPU Use Export Script
TrOCRDecoder float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 1.919 ms 0 - 134 MB NPU TrOCR.onnx
TrOCRDecoder float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 1.354 ms 0 - 61 MB NPU TrOCR.tflite
TrOCRDecoder float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN 1.337 ms 3 - 151 MB NPU Use Export Script
TrOCRDecoder float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 1.546 ms 1 - 149 MB NPU TrOCR.onnx
TrOCRDecoder float Snapdragon X Elite CRD Snapdragon® X Elite QNN 2.224 ms 7 - 7 MB NPU Use Export Script
TrOCRDecoder float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 2.3 ms 68 - 68 MB NPU TrOCR.onnx
TrOCREncoder float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 78.907 ms 0 - 157 MB NPU TrOCR.tflite
TrOCREncoder float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN 73.772 ms 2 - 11 MB NPU Use Export Script
TrOCREncoder float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 53.469 ms 7 - 168 MB NPU TrOCR.tflite
TrOCREncoder float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN 63.778 ms 2 - 148 MB NPU Use Export Script
TrOCREncoder float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 41.135 ms 7 - 24 MB NPU TrOCR.tflite
TrOCREncoder float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN 38.704 ms 2 - 5 MB NPU Use Export Script
TrOCREncoder float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 42.573 ms 7 - 164 MB NPU TrOCR.tflite
TrOCREncoder float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN 39.776 ms 2 - 16 MB NPU Use Export Script
TrOCREncoder float SA7255P ADP Qualcomm® SA7255P TFLITE 78.907 ms 0 - 157 MB NPU TrOCR.tflite
TrOCREncoder float SA7255P ADP Qualcomm® SA7255P QNN 73.772 ms 2 - 11 MB NPU Use Export Script
TrOCREncoder float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 41.334 ms 7 - 24 MB NPU TrOCR.tflite
TrOCREncoder float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN 39.461 ms 2 - 4 MB NPU Use Export Script
TrOCREncoder float SA8295P ADP Qualcomm® SA8295P TFLITE 54.849 ms 7 - 161 MB NPU TrOCR.tflite
TrOCREncoder float SA8295P ADP Qualcomm® SA8295P QNN 51.112 ms 2 - 20 MB NPU Use Export Script
TrOCREncoder float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 41.325 ms 7 - 26 MB NPU TrOCR.tflite
TrOCREncoder float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN 38.716 ms 2 - 4 MB NPU Use Export Script
TrOCREncoder float SA8775P ADP Qualcomm® SA8775P TFLITE 42.573 ms 7 - 164 MB NPU TrOCR.tflite
TrOCREncoder float SA8775P ADP Qualcomm® SA8775P QNN 39.776 ms 2 - 16 MB NPU Use Export Script
TrOCREncoder float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 41.437 ms 7 - 24 MB NPU TrOCR.tflite
TrOCREncoder float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN 39.122 ms 2 - 33 MB NPU Use Export Script
TrOCREncoder float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 40.949 ms 14 - 138 MB NPU TrOCR.onnx
TrOCREncoder float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 33.313 ms 7 - 167 MB NPU TrOCR.tflite
TrOCREncoder float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN 31.253 ms 2 - 151 MB NPU Use Export Script
TrOCREncoder float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 30.905 ms 14 - 169 MB NPU TrOCR.onnx
TrOCREncoder float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 27.077 ms 6 - 164 MB NPU TrOCR.tflite
TrOCREncoder float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN 26.667 ms 2 - 153 MB NPU Use Export Script
TrOCREncoder float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 28.45 ms 12 - 175 MB NPU TrOCR.onnx
TrOCREncoder float Snapdragon X Elite CRD Snapdragon® X Elite QNN 36.496 ms 2 - 2 MB NPU Use Export Script
TrOCREncoder float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 37.527 ms 50 - 50 MB NPU TrOCR.onnx

Installation

Install the package via pip:

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 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.

qai-hub configure --api_token API_TOKEN

Navigate to 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.

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.
python -m qai_hub_models.models.trocr.export

Profiling Results
------------------------------------------------------------
TrOCRDecoder
Device                          : cs_8275 (ANDROID 14)                 
Runtime                         : TFLITE                               
Estimated inference time (ms)   : 4.3                                  
Estimated peak memory usage (MB): [0, 59]                              
Total # Ops                     : 399                                  
Compute Unit(s)                 : npu (399 ops) gpu (0 ops) cpu (0 ops)

------------------------------------------------------------
TrOCREncoder
Device                          : cs_8275 (ANDROID 14)                 
Runtime                         : TFLITE                               
Estimated inference time (ms)   : 78.9                                 
Estimated peak memory usage (MB): [0, 157]                             
Total # Ops                     : 603                                  
Compute Unit(s)                 : npu (603 ops) gpu (0 ops) cpu (0 ops)

How does this work?

This export script leverages Qualcomm® AI Hub 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.

import torch

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

# Load the model
torch_model = Model.from_pretrained()

# Device
device = hub.Device("Samsung Galaxy S24")

# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()

pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])

# Compile model on a specific device
compile_job = hub.submit_compile_job(
    model=pt_model,
    device=device,
    input_specs=torch_model.get_input_spec(),
)

# Get target model to run on-device
target_model = 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.

profile_job = hub.submit_profile_job(
    model=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.

input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
    model=target_model,
    device=device,
    inputs=input_data,
)
    on_device_output = 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.

Deploying compiled model to Android

The models can be deployed using multiple runtimes:

  • TensorFlow Lite (.tflite export): This tutorial provides a guide to deploy the .tflite model in an Android application.

  • QNN (.so export ): This sample app 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. Explore all available models on Qualcomm® AI Hub

License

  • The license for the original implementation of TrOCR can be found here.
  • The license for the compiled assets for on-device deployment can be found here

References

Community