54、记录yolov7 训练、部署ncnn、部署mnn、部署rk3399pro npu、部署openvino、部署oak vpu、部署TensorRT_yolov7-rk3399pro

基本思想：记录一下yolov7训练人头检测和部署oak的vpu相机上和rk3399 pro开发板上，完成需求

ubuntu@ubuntu:~/yolov7$ pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu113

一、准备数据集，来自官方数据集，存在问题，已经修正

链接: https://pan.baidu.com/s/1SUGKRgxeV7ddZpLdILhOwA?pwd=atjt 提取码: atjt

实验数据和模型

链接: https://pan.baidu.com/s/1mjYitnFfGvzLLcCzSQwLbg?pwd=7yaw 提取码: 7yaw

二、在现有的数据集基础上转yolo数据集

ubuntu@ubuntu:~$ git clone https://github.com/WongKinYiu/yolov7.git
Cloning into 'yolov7'...
remote: Enumerating objects: 1094, done.
remote: Total 1094 (delta 0), reused 0 (delta 0), pack-reused 1094
Receiving objects: 100% (1094/1094), 69.89 MiB | 2.21 MiB/s, done.
Resolving deltas: 100% (517/517), done.

下载权重测试一下

ubuntu@ubuntu:~/yolov7/weights$
ubuntu@ubuntu:~/yolov7/w