AXERA-TECH/Helmet-axera

Axera-hed

This version of Axera-hed has been converted to run on the Axera NPU using w8a16 quantization. It is mainly used for detecting whether motor vehicle drivers are wearing helmets in traffic scenarios.

Supported Classes

This model is trained to detect the following 4 classes:

1. 1.helmet
2. 2.head
3. 3.e-bike
4. 4.bike

Compatible with Pulsar2 version: 5.0.

Convert tools links:

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

• —The repo of AXera Platform, where you can get the detailed guide.
• —Pulsar2 Link, How to Convert ONNX to axmodel

Support Platform

https://docs.m5stack.com/zhCN/aihardware/AI_Pyramid-Pro

• —AX650N/AX8850
• —M4N-Dock(爱芯派Pro)
• —AI Pyramid
• —M.2 Accelerator card
• —AX620E
• —爱芯派2
• —Module-LLM

How to use

Download all files from this repository to the device.

python env requirement

pyaxengine

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

wget https://github.com/AXERA-TECH/pyaxengine/releases/download/0.1.3.rc2/axengine-0.1.3-py3-none-any.whl
pip install axengine-0.1.3-py3-none-any.whl

Inference with AX650 Host, such as M4N-Dock(爱芯派Pro)

Input image: [image][image]

run

python3 ax_hed_infer.py --model ./AX650/ax_ax650_hel_algo_V1.0.0.axmodel --img test.jpg

root@ax650:/pcd# python3 ax_hed_infer.py  --model ./AX650/ax_ax650_hel_algo_V1.0.0.axmodel --img test.jpg
[INFO] Available providers:  ['AxEngineExecutionProvider']
[INFO] Using provider: AxEngineExecutionProvider
[INFO] Chip type: ChipType.MC50
[INFO] VNPU type: VNPUType.DISABLED
[INFO] Engine version: 2.10.1s
[INFO] Model type: 2 (triple core)
[INFO] Compiler version: 6.0 79a1e641
Input_name: images, Output_name: ['output0', '167']
Preprocess time: 0.38 ms
Inference time: 17.19 ms
Total detect 2 objects
0: head  0.840 [75.0, 5.0, 110.0, 47.0]
1: bike  0.844 [46.0, 113.0, 143.0, 256.0]

Output image: [image][image]