文章全部YOLOv2源码分析
https://github.com/thtrieu/darkflow tensorflow源码
R-FCN、SSD、YOLO2、faster-rcnn和labelImg实验笔记
http://blog.csdn.net/mmmmmttttff/article/details/53243414深度学习目标检测:RCNN,Fast,Faster,YOLO,SSD比较
https://img-blog.csdn.net/20170110191535928?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvaWtlcnBlbmc=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/Center
yolo v2 android ok(源码)
darknet yolo 计算mAP,recall
YOLO V2 的mAP数据测试
http://blog.csdn.net/lucky__ing/article/details/78310510【YOLO学习】召回率(Recall),精确率(Precision),平均正确率(Average_precision(AP) ),交除并(Intersection-over-Union(IoU))
训练:
在darknet-master下执行:
./darknet detector train cfg/personTrain.data cfg/tiny-yolo.cfg tiny-yolo.conv.14
- 1
训练中断:
- 在保存的最新的权重文件的基础上继续训练,执行:
./darknet detector train cfg/personTrain.data cfg/tiny-yolo.cfg backup/tiny-yolo_8000.weights
- 1
- 也可以执行以下命令:
./darknet partial cfg/tiny-yolo.cfg backup/tiny-yolo_8000.weights tiny-yolo_8000.conv.14 14
./darknet detector train cfg/tiny-yolo.cfg tiny-yolo_8000.conv.14 2>1 | tee person_train_log.txt
- 1
- 2
保存训练中间参数结果person_train_log.txt,用于可视化调参。
yolo训练之训练结果评估环节
YOLO模型训练可视化训练过程中的中间参数
http://blog.csdn.net/Allyli0022/article/details/70171655