Hugging Face's logo

qualcomm
/
EasyOCR

Image-to-Text
PyTorch
LiteRT
android
Model card Files
xet
Community
3
EasyOCR / README.md
qaihm-bot's picture
qaihm-bot
v0.34.0
8ecb0f1 verified
preview code
|
raw
history blame
16.4 kB
metadata
library_name: pytorch
license: other
tags:
  - android
pipeline_tag: image-to-text

EasyOCR: Optimized for Mobile Deployment

Ready-to-use OCR with 80+ supported languages and all popular writing scripts

EasyOCR is a machine learning model that can recognize text in images. It supports 80+ supported languages and all popular writing scripts.

This model is an implementation of EasyOCR found here.

This repository provides scripts to run EasyOCR 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: easyocr-small-stage1
    • Input resolution: 384x384
    • Number of parameters (EasyOCRDetector): 20.8M
    • Model size (EasyOCRDetector) (float): 79.2 MB
    • Number of parameters (EasyOCRRecognizer): 3.84M
    • Model size (EasyOCRRecognizer) (float): 14.7 MB
Model Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
EasyOCRDetector float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 275.378 ms 16 - 47 MB NPU EasyOCR.tflite
EasyOCRDetector float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 77.255 ms 16 - 77 MB NPU EasyOCR.tflite
EasyOCRDetector float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN_DLC 77.222 ms 6 - 46 MB NPU EasyOCR.dlc
EasyOCRDetector float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 41.483 ms 11 - 238 MB NPU EasyOCR.tflite
EasyOCRDetector float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN_DLC 38.731 ms 6 - 20 MB NPU EasyOCR.dlc
EasyOCRDetector float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 71.815 ms 16 - 48 MB NPU EasyOCR.tflite
EasyOCRDetector float SA7255P ADP Qualcomm® SA7255P TFLITE 275.378 ms 16 - 47 MB NPU EasyOCR.tflite
EasyOCRDetector float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 41.032 ms 10 - 176 MB NPU EasyOCR.tflite
EasyOCRDetector float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN_DLC 38.241 ms 6 - 22 MB NPU EasyOCR.dlc
EasyOCRDetector float SA8295P ADP Qualcomm® SA8295P TFLITE 78.458 ms 16 - 52 MB NPU EasyOCR.tflite
EasyOCRDetector float SA8295P ADP Qualcomm® SA8295P QNN_DLC 75.171 ms 3 - 42 MB NPU EasyOCR.dlc
EasyOCRDetector float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 40.596 ms 11 - 238 MB NPU EasyOCR.tflite
EasyOCRDetector float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN_DLC 39.498 ms 6 - 22 MB NPU EasyOCR.dlc
EasyOCRDetector float SA8775P ADP Qualcomm® SA8775P TFLITE 71.815 ms 16 - 48 MB NPU EasyOCR.tflite
EasyOCRDetector float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 40.495 ms 10 - 179 MB NPU EasyOCR.tflite
EasyOCRDetector float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN_DLC 38.982 ms 6 - 19 MB NPU EasyOCR.dlc
EasyOCRDetector float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 39.883 ms 41 - 52 MB NPU EasyOCR.onnx
EasyOCRDetector float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 29.857 ms 16 - 76 MB NPU EasyOCR.tflite
EasyOCRDetector float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN_DLC 28.602 ms 6 - 43 MB NPU EasyOCR.dlc
EasyOCRDetector float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 28.788 ms 5 - 44 MB NPU EasyOCR.onnx
EasyOCRDetector float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 29.227 ms 24 - 60 MB NPU EasyOCR.tflite
EasyOCRDetector float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN_DLC 23.563 ms 3 - 40 MB NPU EasyOCR.dlc
EasyOCRDetector float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 23.753 ms 41 - 77 MB NPU EasyOCR.onnx
EasyOCRDetector float Snapdragon X Elite CRD Snapdragon® X Elite QNN_DLC 40.303 ms 27 - 27 MB NPU EasyOCR.dlc
EasyOCRDetector float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 40.544 ms 65 - 65 MB NPU EasyOCR.onnx
EasyOCRRecognizer float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 480.458 ms 8 - 18 MB CPU EasyOCR.tflite
EasyOCRRecognizer float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 137.431 ms 9 - 27 MB CPU EasyOCR.tflite
EasyOCRRecognizer float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN_DLC 37.902 ms 0 - 198 MB NPU EasyOCR.dlc
EasyOCRRecognizer float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 112.808 ms 7 - 10 MB CPU EasyOCR.tflite
EasyOCRRecognizer float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN_DLC 24.989 ms 0 - 100 MB NPU EasyOCR.dlc
EasyOCRRecognizer float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 353.875 ms 11 - 22 MB CPU EasyOCR.tflite
EasyOCRRecognizer float SA7255P ADP Qualcomm® SA7255P TFLITE 480.458 ms 8 - 18 MB CPU EasyOCR.tflite
EasyOCRRecognizer float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 107.022 ms 8 - 11 MB CPU EasyOCR.tflite
EasyOCRRecognizer float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN_DLC 25.085 ms 0 - 104 MB NPU EasyOCR.dlc
EasyOCRRecognizer float SA8295P ADP Qualcomm® SA8295P TFLITE 203.189 ms 10 - 28 MB CPU EasyOCR.tflite
EasyOCRRecognizer float SA8295P ADP Qualcomm® SA8295P QNN_DLC 40.492 ms 0 - 198 MB NPU EasyOCR.dlc
EasyOCRRecognizer float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 110.1 ms 8 - 10 MB CPU EasyOCR.tflite
EasyOCRRecognizer float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN_DLC 25.287 ms 0 - 102 MB NPU EasyOCR.dlc
EasyOCRRecognizer float SA8775P ADP Qualcomm® SA8775P TFLITE 353.875 ms 11 - 22 MB CPU EasyOCR.tflite
EasyOCRRecognizer float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 115.378 ms 5 - 8 MB CPU EasyOCR.tflite
EasyOCRRecognizer float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN_DLC 25.199 ms 0 - 102 MB NPU EasyOCR.dlc
EasyOCRRecognizer float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 22.099 ms 3 - 10 MB NPU EasyOCR.onnx
EasyOCRRecognizer float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 99.918 ms 9 - 29 MB CPU EasyOCR.tflite
EasyOCRRecognizer float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN_DLC 18.404 ms 0 - 500 MB NPU EasyOCR.dlc
EasyOCRRecognizer float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 15.993 ms 3 - 23 MB NPU EasyOCR.onnx
EasyOCRRecognizer float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 100.743 ms 20 - 35 MB CPU EasyOCR.tflite
EasyOCRRecognizer float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN_DLC 18.95 ms 0 - 504 MB NPU EasyOCR.dlc
EasyOCRRecognizer float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 14.455 ms 3 - 21 MB NPU EasyOCR.onnx
EasyOCRRecognizer float Snapdragon X Elite CRD Snapdragon® X Elite QNN_DLC 25.134 ms 83 - 83 MB NPU EasyOCR.dlc
EasyOCRRecognizer float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 20.59 ms 0 - 0 MB NPU EasyOCR.onnx

Installation

Install the package via pip:

pip install "qai-hub-models[easyocr]"

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.easyocr.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.easyocr.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.easyocr.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.easyocr 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 EasyOCR's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

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

References

Community