justchenhao / sadl_cd Goto Github PK

Thanks to the author for sharing the code and data. The link of Baidu Cloud to share the pretrained data is currently unavailable. Would the author mind sharing it once more? Thanks again.

最近读了您的论文“SemiSANet: A Semi-Supervised High-Resolution Remote Sensing Image Change Detection Model Using Siamese Networks with Graph Attention”，受益匪浅。如果可能的话，我想向您寻求一份该代码的副本，以便更好地探索和理解您的研究成果。
如果您能够提供该代码或相关信息，我将不胜感激。
谢谢您的时间和帮助，期待您的回复。

I use the LEVIR-CD dataset and change the size of the input image to 1024. Why does it report the following error when running?

File "main_ssl.py", line 98, in
train(args)
File "main_ssl.py", line 20, in train
model.train_models()
File "E:\Code\SaDL_CD-main\self_sup\ssl_trainer.py", line 221, in train_models
batch = self.model.on_after_batch_transfer(batch)
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "E:\Code\SaDL_CD-main\self_sup\masksiamsim.py", line 833, in on_after_batch_transfer
x1_t1, x2_t1, m1_t1, m2_t1, pts1, pts2 = self.aug(x_t1, m_t1)
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\torch\nn\modules\module.py", line 1130, in call_impl
return forward_call(*input, **kwargs)
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "E:\Code\SaDL_CD-main\datasets\transforms_sample.py", line 163, in forward
x1, _ = self.augs1.inverse(x1, m1)
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\kornia\augmentation\container\augment.py", line 269, in inverse
input = ApplyInverse.inverse_by_key(input, module, param, dcate, extra_args=extra_args)
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\kornia\augmentation\container\utils.py", line 639, in inverse_by_key
return func.inverse(input, module, param, extra_args=extra_args)
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\kornia\augmentation\container\utils.py", line 228, in inverse
input = module.inverse(
File "D:\Anaconda3\InstallRoad\envs\pytorch\lib\site-packages\kornia\augmentation_2d\geometric\base.py", line 85, in inverse
in_tensor[params['batch_prob']], params=params, flags=flags
IndexError: The shape of the mask [2] at index 0 does not match the shape of the indexed tensor [1, 3, 1024, 1024] at index 0

Can you please disclose your WHU dataset? I am very interested in this 6096/762/762 partition plan and would like to receive your reply. My email is [email protected] Thank you

Hello, may I inquire if there is code available for fine-tuning a pre-trained model on downstream tasks? How should we go about using the pre-trained model for fine-tuning?