speedinghzl / CCNet

CCNet: Criss-Cross Attention for Semantic Segmentation (TPAMI 2020 & ICCV 2019).

speedinghzl/CCNet Issues

Why didn't add self.INF(m_batchsize, height, width) to energy_W
Updated a month ago
Why CCA module do not need position encoding for every single pixels?
Updated 6 months ago
模型参数量
Updated a year ago
loss to nan When applied to my segmentation network
Closed 4 years ago1
Inplace-ABN installation - which version of Inplace-ABN you installed?
Updated a year ago
some issues when training in yolov5 + CrissCrossAttention
Updated a year ago
The attention of shape `(H*W) x (H+W-1)` cannot be seen in the code variables?
Updated 2 years ago
关于实现问题
Updated 2 years ago2
Inplace-ABN does not work
Updated 2 years ago
about INF
Updated 2 years ago1
cc_attention def INF
Updated 2 years ago3
Why self.gamma is intialized to zero
Updated 2 years ago2
About image classification tasks, compared with nolocal
Updated 2 years ago2
Some questions about 'Criss-Cross Attention'
Updated 3 years ago2
Question
Updated 3 years ago
Regarding RCCA
Updated 3 years ago1
cc_attention for conv1d?
Updated 3 years ago
Can you offer me a trained models using pytorch 1.5+?
Updated 4 years ago
将RCCA应用到视频任务中，loss没有完全收敛
Updated 4 years ago1
I check your implementation for 3d in pure-python branch
Updated 4 years ago
No code implementation about discriminative loss ?
Updated 4 years ago2
Will the pure pytorch architecture support the 3D Criss Cross / temporal?
Updated 4 years ago
resnet101-imagenet.pth. doesn't work
Updated 4 years ago1
question
Updated 4 years ago1
padding error
Closed 4 years ago
RuntimeError: Error building extension 'rcca'"
Updated 4 years ago2
pytorch1.1的版本在pytorch1.6 cuda10.0的环境下安装报错
Closed 4 years ago1
Error while running run_local.sh
Updated 4 years ago3
rcca compile error #7
Closed 4 years ago2
RuntimeError: CUDA Error encountered in <function Lib.bn_mean_var_cuda at 0x7f96e5e9d2f0>
Updated 4 years ago2
Using CCNet with Pytorch>= 1.5
Updated 4 years ago3
ImportError: libcudart.so.9.2: cannot open shared object file: No such file or directory
Updated 4 years ago3
Figure 6 Code
Updated 4 years ago2
Training speed slower than Non-local attention
Closed 4 years ago2
Getting very low mIoU on Cityscapes
Updated 4 years ago3
AssertionError: Default process group is not initialized
Closed 4 years ago4
Could not run the code
Updated 4 years ago1
about the detail of cc-attention mathematical derivation
Closed 4 years ago2
Error evualuating with only one GPU
Closed 4 years ago2
关于数据集的问题
Updated 4 years ago1
1.5.1 Detected CUDA files, patching ldflags Emitting ninja build file /home/ices/work/sxc/cc_attention/build/build.ninja... Building extension module rcca... ninja: no work to do. Loading extension module rcca... Traceback (most recent call last): File "/home/ices/work/sxc/cc_attention/functions.py", line 121, in <module> out = ca(x, y, z) File "/home/ices/anaconda3/envs/pytorch_1_2_yzz/lib/python3.6/site-packages/torch/nn/modules/module.py", line 547, in __call__ result = self.forward(*input, **kwargs) TypeError: forward() takes 2 positional arguments but 4 were given
Updated 4 years ago
Can you provide the pretrained Resnet50-RCCA(2) model? Thanks!
Updated 4 years ago
ADE20K
Updated 4 years ago
Memory usage
Updated 4 years ago1
about attention map visual
Updated 4 years ago2
Hope to support mixed-precision training
Updated 4 years ago2
About ccnet.py. 'return [x, x_dsn]'
Updated 4 years ago1
可视化
Updated 4 years ago1
关于多尺度预测
Updated 4 years ago
test.py to support pytorch 1.1
Updated 5 years ago