Person_car-axera

Axera-PCD

This version of Axera-PCD has been converted to run on the Axera NPU using w8a16 quantization. It is mainly used for person and car detection in conventional scenarios.

Supported Classes

This model is trained to detect the following 3 classes:

1. person
2. car
3. person_cycle

Compatible with Pulsar2 version: 6.0.

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/zh_CN/ai_hardware/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:

run

python3 ax_pcd_infer.py --model ./ax_ax650_pcd_tiny_algo_V2.0.0.axmodel --img car_away_1920x1080.jpg

root@ax650:/pcd# python3 ax_pcd_infer.py  --model ./AX650/ax_ax650_pcd_tiny_algo_V2.0.0.axmodel --img car_away_1920x1080.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 ec0bb1bc
Input_name: images, Output_name: ['/model.19/Concat_output_0', '/model.19/Concat_1_output_0']
Preprocess time: 8.79 ms
Inference time: 19.45 ms
Postprocess time: 5.38 ms
Total detect 15 objects
0: person        0.644 [1320.0, 614.0, 1357.0, 729.0]
1: person        0.644 [1726.0, 769.0, 1794.0, 924.0]
2: car   0.889 [182.0, 432.0, 251.0, 483.0]
3: car   0.889 [333.0, 553.0, 444.0, 657.0]
4: car   0.889 [110.0, 677.0, 266.0, 809.0]
5: car   0.889 [663.0, 811.0, 860.0, 971.0]
6: car   0.889 [25.0, 920.0, 279.0, 1078.0]
7: car   0.644 [125.0, 336.0, 170.0, 370.0]
8: car   0.644 [264.0, 401.0, 330.0, 453.0]
9: car   0.644 [496.0, 496.0, 595.0, 564.0]
10: car  0.644 [447.0, 530.0, 564.0, 620.0]
11: car  0.644 [377.0, 668.0, 531.0, 800.0]
12: car  0.644 [425.0, 808.0, 643.0, 1044.0]
13: car  0.250 [453.0, 401.0, 510.0, 444.0]
14: car  0.250 [785.0, 1007.0, 1095.0, 1080.0]
Result saved to ./out.jpg

Output image: