I don't know if my partner is training his own YOLOv5 data set, if the training is interrupted, how to restore the original data progress? Let's look at the resolution process.
1. At this time, the data set is being trained normally, with 100 rounds of training, and the training is in progress:
2. Sudden computer restart, open it again and find that 'runs-train-exp7' in the root directory has trained 0-2 rounds, interrupted and 2 rounds:
3. At this time, modify the default value under the '--resume' parameter, and change False to True:
parser.add_argument('--resume', nargs='?', const=True, default=True, help='resume most recent training')
4. YOLOv5 has its own breakpoint save, breakpoint resume so that it can resume training, as shown in the figure:
Training is not easy, I hope everyone can train their ideal model!
Refer to the blog: https://blog.csdn.net/weixin_45800887/article/details/124222937