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 (TrOCRDecoder): 38.3M
- Model size (TrOCRDecoder) (float): 146 MB
- Number of parameters (TrOCREncoder): 23.0M
- Model size (TrOCREncoder) (float): 87.8 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 | 4.34 ms | 0 - 82 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 4.142 ms | 5 - 72 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 2.429 ms | 0 - 144 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 2.747 ms | 7 - 133 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 2.066 ms | 0 - 465 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 2.023 ms | 2 - 27 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.902 ms | 0 - 82 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.847 ms | 0 - 64 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 4.34 ms | 0 - 82 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 4.142 ms | 5 - 72 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 2.081 ms | 0 - 504 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 2.027 ms | 1 - 25 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 2.914 ms | 0 - 75 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 2.902 ms | 0 - 58 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 2.084 ms | 0 - 515 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 2.054 ms | 3 - 33 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.902 ms | 0 - 82 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 2.847 ms | 0 - 64 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 2.062 ms | 0 - 504 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 2.021 ms | 2 - 25 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 2.768 ms | 0 - 176 MB | NPU | TrOCR.onnx |
TrOCRDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.461 ms | 0 - 159 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 1.484 ms | 0 - 145 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 2.011 ms | 0 - 142 MB | NPU | TrOCR.onnx |
TrOCRDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 1.384 ms | 0 - 70 MB | NPU | TrOCR.tflite |
TrOCRDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 1.344 ms | 2 - 156 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 1.686 ms | 2 - 160 MB | NPU | TrOCR.onnx |
TrOCRDecoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 2.252 ms | 673 - 673 MB | NPU | TrOCR.dlc |
TrOCRDecoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2.39 ms | 68 - 68 MB | NPU | TrOCR.onnx |
TrOCREncoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 77.233 ms | 7 - 167 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 73.626 ms | 2 - 152 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 50.044 ms | 7 - 170 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 71.235 ms | 2 - 153 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 41.575 ms | 7 - 27 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 39.479 ms | 2 - 38 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 42.469 ms | 7 - 167 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 40.352 ms | 2 - 151 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 77.233 ms | 7 - 167 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 73.626 ms | 2 - 152 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 40.559 ms | 7 - 27 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 39.68 ms | 2 - 38 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 53.503 ms | 7 - 165 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 50.849 ms | 2 - 151 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 41.346 ms | 8 - 27 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 39.47 ms | 2 - 36 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 42.469 ms | 7 - 167 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 40.352 ms | 2 - 151 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 40.719 ms | 7 - 30 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 39.703 ms | 2 - 43 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 39.56 ms | 16 - 52 MB | NPU | TrOCR.onnx |
TrOCREncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 32.916 ms | 6 - 170 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 32.248 ms | 82 - 235 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 32.449 ms | 15 - 176 MB | NPU | TrOCR.onnx |
TrOCREncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 28.297 ms | 6 - 166 MB | NPU | TrOCR.tflite |
TrOCREncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 26.788 ms | 2 - 161 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 25.626 ms | 15 - 200 MB | NPU | TrOCR.onnx |
TrOCREncoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 38.141 ms | 84 - 84 MB | NPU | TrOCR.dlc |
TrOCREncoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 37.091 ms | 51 - 51 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
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
- TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models
- Source Model Implementation
Community
- Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
- For questions or feedback please reach out to us.