wolf2345的博客

环境要求： 1、ubuntu16.04或者ubuntu18.04 2、cmake_minimum_required(VERSION 3.12)（升级包见software） 升级cmake： （1）tar zxvf cmake-3.12.2-Linux-x86_64.tar.gz （2）mv cmake-3.12.2-Linux-x86_64 /opt/cmake-3.12.2 （3）ln -sf /opt/cmake-3.12.2/bin/* /usr/bin/ 3、CUDA>=10..

1、训练yolov4 ./darknet detector train /home/pc/Documents/yolov4/data/0827/code/obj.data /home/pc/Documents/yolov4/data/0827/code/yolov4-obj.cfg /home/pc/Documents/yolov4/data/0827/code/yolov4.conv.137 -map 2、测试打印map ./darknet detector test /home/pc/Docume

1、网络模型 主要修改classes和其上一层网络层输出（classes+5）*3->(classes+[xmin,ymin,xmax,ymax,c])*Scalar lr_rate和max_batches自行配置，建议初始学习率设置为0.001，max_batches=classes*2000 [net] # Testing #batch=1 #subdivisions=1 # Training batch=96 subdivisions=32 width=416 height=416 cha

1、图片重命名->rename.py import os #import cv2 import shutil jpg_path = '/media/em/data_1/0716/data2/socket/mark' xml_path = '/media/em/data_1/0716/data2/socket/xml/' out_jpg_path = '/media/em/data_1/0716/data2/socket/JPEGImages1/' out_xml_path = '/media/

Done! Loaded 162 layers from weights-file Learning Rate: 0.001, Momentum: 0.949, Decay: 0.0005 If error occurs - run training with flag: -dont_show Saving weights to /home/pc/Documents/yolov4/darknet-master/backup/backup20200702/yolo-obj_final.weights Cr..