NAIC_Person_ReID_DMT
NAIC_Person_ReID_DMT copied to clipboard
heshuting555
test集pids都设为1，训练过程中的accuracy是怎么得来的呢
训练的acc是针对训练过程的label呢，测试过程因为没有ground truth所以都置为1。
发自我的iPhone
------------------ 原始邮件 ------------------ 发件人: zhangting <[email protected]> 发送时间: 2020年9月27日 16:10 收件人: heshuting555/NAIC_Person_ReID_DMT <[email protected]> 抄送: Subscribed <[email protected]> 主题: 回复：[heshuting555/NAIC_Person_ReID_DMT] test集pids都设为1，训练过程中的accuracy是怎么得来的呢 (#7)
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.
=> NAIC Competition data loaded Dataset statistics:
subset | # ids | # images | # cameras
train | 15506 | 68672 | 1 query | 1 | 2900 | 1 gallery | 1 | 40466 | 1
using resnet101_ibn_a as a backbone
Loading pretrained ImageNet model......from ./imagenet_models/resnet101_ibn_a.pth.tar
using arcface with s:30, m: 0.4
Traceback (most recent call last):
File "test.py", line 44, in
=> NAIC Competition data loaded
Dataset statistics:
subset | # ids | # images | # cameras
train | 15506 | 68672 | 1
query | 1 | 2900 | 1 gallery | 1 | 40466 | 1
using resnet101_ibn_a as a backbone Loading pretrained ImageNet model......from ./imagenet_models/resnet101_ibn_a.pth.tar using arcface with s:30, m: 0.4 Traceback (most recent call last): File "test.py", line 44, in model.load_param(cfg.TEST.WEIGHT) File "/data/zhangting/NAIC_Person_ReID_DMT/model/make_model.py", line 157, in load_param self.state_dict()[i].copy_(param_dict[i]) KeyError: 'module.base.conv1.weight' 请问这个报错是怎么回事呢
遇到了同样的错误，请问解决了吗
load的模型是多卡训练的模型，改成self.state_dict()[i.replace('module.', '')].copy_(param_dict[i]) 即可