wlsh
wlsh
同样的问题已经解决. 搜索的是docker引起的莫名其妙的错误. 可能是我用的集群分给给每个用户是docker虚拟出来的 解决方法： https://github.com/whai362/PSENet/blob/4d95395658662f2223805c36dcd573d9e190ce26/train.py#L158 改为 **num_workers=0** 运行后出现另外一个新错误. UnicodeDecodeError: 'ascii' codec can't decode byte 0xef in position 0: ordinal not in range(128) 卸载旧版本mmcv，安装最新版本的mmcv即可 **pip uninstall mmcv** **pip install mmcv** 解决完后，就能正常运行了..然后把num_worker改为gpu_num整数倍也没问题了
Hi, I also encountered this problem. In addition to removing the missing video, have you successfully reproduced it, if so, how mAP that the model get. the missing video list...
ok, I've run through the code, but I haven't got the result yet. could you add my wechat account "wang_leshan" for communicate?