AXERA-TECH/QRCode-axera

QRCode_det

This version of QRCode detetion model has been converted to run on the Axera NPU using w8a16 quantization.

This model has been optimized with the following LoRA:

Compatible with Pulsar2 version: 5.1

Convert tools links:

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

• The original repo, which you can get the detail of guide

• Pulsar2 Link, How to Convert ONNX to axmodel

• The repo of AXera Platform,which you can learn how to compile the C++ demo

Support Platform

• AX650
  • M4N-Dock(爱芯派Pro)
  • M.2 Accelerator card
• AX630C
  • 爱芯派2
  • Module-LLM
  • LLM630 Compute Kit
• AX637

Chips	model	cost
	yolov5n	1.73 ms
	yolov8n	3.64 ms
	yolov9t	4.75 ms
AX650	yolov10n	3.67 ms
	yolo11n	3.42 ms
	yolo12n	6.87 ms
	yolo26n	3.24 ms
	NanodetPlus	2.16 ms
	DEIMv2_femto(u16)	3.76 ms
	yolov5n	5.79 ms
	yolov8n	9.26 ms
	yolov9t	11.6 ms
AX630C	yolov10n	9.71 ms
	yolo11n	9.65 ms
	yolo12n	20.24 ms
	yolo26n	10.04 ms
	NanodetPlus	5.93 ms
	yolov5n	2.11 ms
	yolov8n	4.04 ms
	yolov9t	4.91 ms
AX637	yolov10n	4.05 ms
	yolo11n	3.84 ms
	yolo12n	6.40 ms
	yolo26n	3.50 ms
	NanodetPlus	2.38 ms

How to use

Download all files from this repository to the device

.
├── config.json
├── CPP
│   ├── ax_deimv2_qrcode_batch
│   ├── ax_nanodetplus_qrcode_batch
│   ├── ax_yolov5_qrcode_batch
│   ├── ax_yolov8_qrcode_batch
│   └── ax_yolo26_qrcode_batch
├── cpp_result.png
├── images
│   ├── qrcode_01.jpg
│   ├── qrcode_02.jpg
│   ├── qrcode_03.jpg
|   ├── ...
│   └── qrcode_55.jpg
├── model
│   ├── AX620E
│   │   ├── nanodet-plus-m_630_npu1.axmodel
│   │   ├── yolo11n_630_npu1.axmodel
│   │   ├── yolo12n_630_npu1.axmodel
│   │   ├── yolo26n_630_npu1.axmodel
│   │   ├── yolov10n_630_npu1.axmodel
│   │   ├── yolov5n_630_npu1.axmodel
│   │   ├── yolov8n_630_npu1.axmodel
│   │   └── yolov9t_630_npu1.axmodel
│   ├── AX637
│   │   ├── nanodet-plus-m_637_npu1.axmodel
│   │   ├── yolo11n_637_npu1.axmodel
│   │   ├── yolo12n_637_npu1.axmodel
│   │   ├── yolo26n_637_npu1.axmodel
│   │   ├── yolov10n_637_npu1.axmodel
│   │   ├── yolov5n_637_npu1.axmodel
│   │   ├── yolov8n_637_npu1.axmodel
│   │   └── yolov9t_637_npu1.axmodel
│   └── AX650
│       ├── deimv2_femto_650_npu1_u16.axmodel
│       ├── nanodet-plus-m_650_npu1.axmodel
│       ├── yolo11n_650_npu1.axmodel
│       ├── yolo12n_650_npu1.axmodel
│       ├── yolo26n_650_npu1.axmodel
│       ├── yolov10n_650_npu1.axmodel
│       ├── yolov5n_650_npu1.axmodel
│       ├── yolov8n_650_npu1.axmodel
│       └── yolov9t_650_npu1.axmodel
├── py_result.png
├── python
│   ├── QRCode_axmodel_infer_DEIMv2.py
│   ├── QRCode_axmodel_infer_Nanodet.py
│   ├── QRCode_axmodel_infer_v5.py
│   ├── QRCode_axmodel_infer_v8.py
│   ├── QRCode_axmodel_infer_26.py
│   ├── QRCode_onnx_infer_DEIMv2.py
│   ├── QRCode_onnx_infer_Nanodet.py
│   ├── QRCode_onnx_infer_v5.py
│   ├── QRCode_onnx_infer_v8.py
│   ├── QRCode_onnx_infer_26.py
│   └── requirements.txt
└── README.md

Inference

Input Data:

|-- images
|   `-- qrcode_01.jpg
|   `-- qrcode_02.jpg
|   `-- qrcode_03.jpg
|   `-- qrcode_04.jpg...

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

Python

run with python3 QRCode_axmodel_infer_xxx.py

root@ax650:~/QRCode# python3 QRCode_axmodel_infer_DEIMv2.py
[INFO] Available providers:  ['AxEngineExecutionProvider']
[INFO] Using provider: AxEngineExecutionProvider
[INFO] Chip type: ChipType.MC50
[INFO] VNPU type: VNPUType.DISABLED
[INFO] Engine version: 2.12.0s
[INFO] Model type: 2 (triple core)
[INFO] Compiler version: 4.2 b98901c3
识别成功！
图片 ./qrcode_test/qrcode_01.jpg 处理耗时: 0.2165 秒
识别成功！
图片 ./qrcode_test/qrcode_02.jpg 处理耗时: 0.1540 秒
识别成功！
图片 ./qrcode_test/qrcode_03.jpg 处理耗时: 0.1456 秒
识别成功！
图片 ./qrcode_test/qrcode_05.jpg 处理耗时: 0.1449 秒

Output:

C++

./ax_xxx_qrcode_batch -m xxx_npu1.axmodel -i images/

Output: