ming71 / cfc-net Goto Github PK

hello，I want to get some information of dateset your project used. I use DOTA and there are any error.

Please can you give us , steps to prepare data format needed to train NWPU VHR dataset ?

您好，请问您是如何可视化特征的，像论文中的那样，在代码的哪个部分呢，想学习一下

请问你可以把DOTA的trainval.txt的文件发给我一份吗，我的邮箱[email protected]

非常感谢您在在此方向的贡献。最近看到您的论文，非常精彩，但是我发现现在的代码不完整和没有训练好模型，请问什么时候可以公布

我用hrsc训练的时候，通道注意力的

def forward(self, x):
avg_pool = F.avg_pool2d(x, x.size(2), stride=x.size(2))# 平均池化操作
res=self.gate_c(avg_pool).unsqueeze(2).unsqueeze(3).expand_as(x)# 通过通道注意力模块处理平均池化的结果
return res
平均池化报错了，RuntimeError: Given input size: (256x100x96). Calculated output size: (256x1x0). Output size is too small
因该是部分图片大小不太规矩，请问您当时是怎么解决的呢？

Hello,can you give a weight file?You can give the file,Thanks,my Email:[email protected]

torch_warmup_lr这个包在哪里

在fpn中的
def forward(self, x):
avg_pool = F.avg_pool2d(x, x.size(2), stride=x.size(2))
return self.gate_c(avg_pool).unsqueeze(2).unsqueeze(3).expand_as(x)
出现报错，请问如何解决啊

 Epoch   gpu_mem       ref       cls       reg     total   targets  img_size

0% 0/260 [00:00<?, ?it/s]/usr/local/lib/python3.7/dist-packages/torch/cuda/memory.py:393: FutureWarning: torch.cuda.memory_cached has been renamed to torch.cuda.memory_reserved
FutureWarning)
0/100 2.42G 1.26 2.07 1.18 4.5 6 416: 15% 40/260 [00:24<02:13, 1.65it/s]
Traceback (most recent call last):
File "/content/drive/MyDrive/CFC-Net/train.py", line 316, in
train_model(arg, hyps)
File "/content/drive/MyDrive/CFC-Net/train.py", line 160, in train_model
losses = model(ims, gt_boxes,process =epoch/epochs )
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/content/drive/MyDrive/CFC-Net/models/model.py", line 102, in forward
cls_feats = self.cls_attention(features, process)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/content/drive/MyDrive/CFC-Net/models/fpn.py", line 305, in forward
atten_ft = self.channel_att(fp) * self.spatial_att(fp)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/content/drive/MyDrive/CFC-Net/models/fpn.py", line 265, in forward
return self.gate_c(avg_pool).unsqueeze(2).unsqueeze(3).expand_as(x)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/container.py", line 139, in forward
input = module(input)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/linear.py", line 114, in forward
return F.linear(input, self.weight, self.bias)
RuntimeError: mat1 and mat2 shapes cannot be multiplied (2x512 and 256x16)

在losses.py里 max_pos, armmax_pos = pos.max(0)
RuntimeError: cannot perform reduction function max on tensor with no elements because the operation does not have an identity
刚开始以为是batch_size的问题，以为解决了，但是后续的Epoch还是会报错，到底是哪里的问题呢

Is there any available pretrained model/weights for this repo?

验证时候遇到“RuntimeError:Given input size :(2565652).Calculated output size:(25610).Output size is too small”问题，请问怎么解决？是什么问题

该行代码不适合多类别的使用的，会索引越界
cls_targets[positive_indices, assigned_annotations[positive_indices, -1].long()] = 1

拜读了大佬的神作，想问一下什么时候会公布全部的源码啊？

Hello,I want to know CFC-Net what is the epoch