satrn

Convert tools links:

For those who are interested in model conversion, you can try to export onnx or axmodel through

Installation

conda create -n open-mmlab python=3.8 pytorch=1.10 cudatoolkit=11.3 torchvision -c pytorch -y
conda activate open-mmlab
pip3 install openmim
git clone https://github.com/open-mmlab/mmocr.git
cd mmocr
mim install -e .

Support Platform

AX650
- M4N-Dock(爱芯派Pro)
- M.2 Accelerator card

The speed measurements(under different NPU configurations ) of the two parts of SATRN:

(1) backbone+encoder

(2) decoder

	backbone+encoder(ms)	decoder(ms)
NPU1	20.494	2.648
NPU2	9.785	1.504
NPU3	6.085	1.384

How to use

Download all files from this repository to the device

.
├── axmodel
│   ├── backbone_encoder.axmodel
│   └── decoder.axmodel
├── demo_text_recog.jpg
├── onnx
│   ├── satrn_backbone_encoder.onnx
│   └── satrn_decoder_sim.onnx
├── README.md
├── run_axmodel.py
├── run_model.py
└── run_onnx.py

python env requirement

1. pyaxengine

https://github.com/AXERA-TECH/pyaxengine

wget https://github.com/AXERA-TECH/pyaxengine/releases/download/0.1.1rc0/axengine-0.1.1-py3-none-any.whl
pip install axengine-0.1.1-py3-none-any.whl

2. satrn

satrn installation

Inference onnxmodel

python run_onnx.py

input:

output:

pred_text: STAR
score: [0.9384028315544128, 0.9574984908103943, 0.9993689656257629, 0.9994958639144897]

Inference with AX650 Host

check the reference for more information