Skip to content

mywang44/YOLOv1_QRcode_Detection

Repository files navigation

YOLOv1-QRcode

QRcode detection project using LNN(linger&thinker)

介绍

本仓库利用LNN工具链实现二维码检测模型的落地。主要包括浮点训练、量化训练、模型打包、模拟引擎执行、固件烧录并芯片运行。其中固件烧录并芯片运行需要在聆思的开发板上来完成。

环境配置

linger环境配置及安装

https://github.com/LISTENAI/linger/blob/main/doc/tutorial/install.md

thinker环境配置及安装

https://github.com/LISTENAI/thinker/blob/main/thinker/docs/tutorial/install.md

requirement

数据集

二维码数据下载链接:https://pan.baidu.com/s/1d66HKN-8773D2FmvyhMwFg

主要流程

模型训练

运行脚本train.py, 浮点训练(float)、约束训练(clamp)、量化训练(quant)会按顺序一次性执行
最终该脚本会在./tmp.ignore/文件夹下生成一个YOLO.quant.onnx

模型打包

切换到thinker-env环境,使用thinker离线工具tpacker将刚才生成的onnx计算图打包

tpacker -g ./tmp.ignore/YOLO.quant.onnx -d True -o ./test_thinker/model.bin

推理执行

使用调用示例工程test_thinker,指定输入数据、资源文件和输出文件夹名称即可运行模拟代码。

chmod +x ./bin/test_thinker  
./bin/test_thinker ./demo/test_thinker/0000964.bin ./demo/test_thinker/model.bin output.bin 3 64 64 0

模型评估

float模式: ---class qr_code AP 0.9584357337014008---
quant模式: ---class qr_code AP 0.9554898401324852---

模型部署

在CSK6多模态芯片上部署运行该算法模型:https://cloud.listenai.com/CSKG962172/duomotai_ap.git
QRcode detection on CSK6

致谢

代码参考: https://github.com/abeardear/pytorch-YOLO-v1/tree/master

About

QRcode detection project using LNN(linger&thinker)

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published
pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy