easyocr_ov_benchmark

Scripts for benchmarking of OpenVINO integration into EasyOCR
This repo helps to benchmark OpenVINO integration into EasyOCR and compare it with an original one based on Pytorch (on CPU).

Installation

1. Create a virtual environment or use an existing one: On Linux:

python3 -m venv env
source env/bin/activate

On Windows:

python -m venv env
env\Scripts\activate

2. Clone the repo:

git clone https://github.com/avbelova/easyocr_ov_benchmark.git
cd easyocr_ov_benchmark

4. Install OpenVINO integration into EasyOCR:

pip install git+https://github.com/avbelova/EasyOCR.git

Run benchmarking

python easyocr_ov_benchmark.py <PATH-TO-AN-INPUT-IMAGE> <LANGUAGE>

Additianally you can specify an inference device for OpenVINO (default one is CPU) and a number of iterations (default is 10).
For example the following command runs 20 benchmarking iterations of English language recognition on Intel(R) Processor Graphics:

python easyocr_ov_benchmark.py EasyOCR/examples/english.png en -n 20 -d ov_gpu

Running Chinese+English text recognition benchmarking on Intel(R) Discrete Graphics:

python easyocr_ov_benchmark.py EasyOCR/examples/chinese.jpg ch_sim en -d ov_gpu.1