about data2

[chengyu89527]

I check the google drive link , there are 3 file (image,label,visual) .which one is stand for infrared image. I find none of them is infrared

[Linfeng-Tang]

In fact, the images in the images folder contain four channels, namely three channels of visible images and a single channel of infrared images, and you need to separate them out. In addition, you can use the MSRS dataset (https://github.com/Linfeng-Tang/MSRS) to train and test SeAFusion.

[chengyu89527]

Thanks a lot, by the way, we need build log folder before braining and evalution

[chengyu89527]

also evalute.py gives : Traceback (most recent call last): File "evaluate.py", line 191, in evaluate() File "evaluate.py", line 165, in evaluate net.load_state_dict(torch.load(save_pth)) File "/home/chengyu/anaconda3/envs/syn38/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1052, in load_state_dict self.class.name, "\n\t".join(error_msgs))) RuntimeError: Error(s) in loading state_dict for BiSeNet: size mismatch for conv_out.conv_out.weight: copying a param with shape torch.Size([9, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([19, 128, 1, 1]). size mismatch for conv_out16.conv_out.weight: copying a param with shape torch.Size([9, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([19, 64, 1, 1]).

[chengyu89527]

another issue is setup and restore model /data/pcl/datasets/cityscapes/Fusion/val Traceback (most recent call last): File "evaluate.py", line 191, in evaluate() File "evaluate.py", line 172, in evaluate dsval = CityScapes(dspth, mode='val') File "/home/chengyu/SeAFusion-main/cityscapes.py", line 43, in init folders = os.listdir(impth) FileNotFoundError: [Errno 2] No such file or directory: '/data/pcl/datasets/cityscapes/Fusion/val' where I can find the the val dataset

[Linfeng-Tang]

In order to run evaluate. py successfully， you need to rewrite the dataloader according to how you store your datasets.

[chengyu89527]

thanks，I need generate some new fusion data 

成大宇 @.***

 

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2023年4月27日(星期四) 下午4:20 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [Linfeng-Tang/SeAFusion] about data2 (Issue #13)

In order to run evaluate. py successfully， you need to rewrite the dataloader according to how you store your datasets.

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

[Linfeng-Tang]

We have updated evaluate.py and added the relevant data to run evaluate.py.