2021/5/20 voc测试数据集 yolov4训练

1 train 结束 done，生成多个文件。

结果 ：

[2021-05-20 21:37:14,552]-[train.py line:279]:===== Validate ===== 
[2021-05-20 21:37:14,552]-[train.py line:281]:val img size is 416 
100%|██████████| 49/49 [00:03<00:00, 13.99it/s]
[2021-05-20 21:37:18,852]-[train.py line:287]:insulator --> mAP : 0.794868071332103 
[2021-05-20 21:37:18,852]-[train.py line:287]:knife-2 --> mAP : 0.9553823953823954 
[2021-05-20 21:37:18,852]-[train.py line:290]:mAP : 0.8751252333572492 
[2021-05-20 21:37:18,852]-[train.py line:292]:inference time: 345.29 ms 
[2021-05-20 21:37:19,204]-[train.py line:296]:save weights done 
[2021-05-20 21:37:19,204]-[train.py line:297]:  ===test mAP:0.875 
[2021-05-20 21:37:19,204]-[train.py line:313]:  ===cost time:42.5396s 
[2021-05-20 21:37:19,204]-[train.py line:316]:=====Training Finished.   best_test_mAP:0.899%==== 

Process finished with exit code 0

生成权重文件  （前四个）：

2

 1）下一步  检测

找到 data-test 代码，修改，运行 evl-voc.py，显示图片。

2） 最终 画图  ，得到图表。

3 另外，由于明天需要汇报，需要完成 论文 开头前俩段，今晚。代码训练阶段完成，下一步检测 detect。