AbnerHqC / GaitSet

A flexible, effective and fast cross-view gait recognition network
577 stars 170 forks source link

提供的iCASIA的权重文件,测试的结果比论文中给的结果好很多,特别是对于CL #114

Open shanglidan opened 4 years ago

shanglidan commented 4 years ago

hello,这是我用提供的test.py测试的结果:

论文中: CL | 61.4 | 75.4 | 80.7 | 77.3 | 72.1 | 70.1 | 71.5 | 73.5 | 73.5 | 68.4 | 50.0 自己训,只修改了数据路径(4卡): CL 61.2 | 71.9 | 76.1 | 73.8 | 67.3 | 67.0 | 68.2 | 71.0 | 73.4 | 67.3 | 51.7 提供的权重: CL 69.2 | 78.9 | 79.1 | 76.8 | 69.2 | 67.7 | 70.2 | 73.4 | 74.5 | 70.7 | 58.4

发现,提供的权重比论文中给出的指标要高很多,但是我自己从头开始训的权重,测试的结果比论文中低一些,想请教一下可能的原因。

heheli1998 commented 3 years ago

您好,请问怎么下载提供的权重呢

heheli1998 commented 3 years ago

您好,怎么下载提供的权重呢

YunhaoDu commented 3 years ago

使用作者提供的权重,我和你得到了类似的结果: image

YunhaoDu commented 3 years ago

您好,怎么下载提供的权重呢

work文件夹下

shirenyang commented 2 years ago

为什么我的显示结果都是2.0 2.0 image

zhang123-sys commented 2 years ago

raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format( RuntimeError: Error(s) in loading state_dict for SetNet: Missing key(s) in state_dict: "set_layer1.forward_block.conv.weight", "set_layer2.forward_block.conv.weight", "set_layer3.forward_block.conv.weight", "set_layer4.forward_block.conv.weight", "set_layer5.forward_block.conv.weight", "set_layer6.forward_block.conv.weight", "gl_layer1.conv.weight", "gl_layer2.conv.weight", "gl_layer3.conv.weight", "gl_layer4.conv.weight", "fc_bin.0". Unexpected key(s) in state_dict: "module.set_layer1.forward_block.conv.weight", "module.set_layer2.forward_block.conv.weight", "module.set_layer3.forward_block.conv.weight", "module.set_layer4.forward_block.conv.weight", "module.set_layer5.forward_block.conv.weight", "module.set_layer6.forward_block.conv.weight", "module.gl_layer1.conv.weight", "module.gl_layer2.conv.weight", "module.gl_layer3.conv.weight", "module.gl_layer4.conv.weight", "module.fc_bin.0". image

使用作者提供的权重,我和你得到了类似的结果: image

zhang123-sys commented 2 years ago

您好,可以看我用作者提供的权重测试的报错吗? image

dyhBUPT commented 2 years ago

您好,可以看我用作者提供的权重测试的报错吗? image

多卡模型加载了单卡权重

zhang123-sys commented 2 years ago

谢谢,作者提供的权重是单卡权重?

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2022年9月21日(星期三) 下午2:28 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [AbnerHqC/GaitSet] 提供的iCASIA的权重文件,测试的结果比论文中给的结果好很多,特别是对于CL (#114)

您好,可以看我用作者提供的权重测试的报错吗?

多卡模型加载了单卡权重

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

zhang123-sys commented 2 years ago

RuntimeError: shape '[3840, 32, 4, -1, 2]' is invalid for input of size 491520

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2022年9月21日(星期三) 下午2:28 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [AbnerHqC/GaitSet] 提供的iCASIA的权重文件,测试的结果比论文中给的结果好很多,特别是对于CL (#114)

您好,可以看我用作者提供的权重测试的报错吗?

多卡模型加载了单卡权重

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

zkzzhou commented 1 year ago

为什么我的显示结果都是2.0 2.0 image

请问您解决了嘛