Non-local_pytorch

Non-local_pytorch copied to clipboard

在运行第三点时 $ CUDA_VISIBLE_DEVICES=0,1 python nl_map_save.py 报这个错，该怎么改啊

3

Traceback (most recent call last): File "nl_map_save.py", line 20, in net.load_state_dict(torch.load('weights/net.pth')) File "/home/user/anaconda3/envs/TEST2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1045, in load_state_dict self.__class__.__name__, "\n\t".join(error_msgs))) RuntimeError: Error(s) in loading state_dict for Network: Missing key(s) in state_dict:...

wys2929

About GPU memory usage

3

If non-local is applied to the low-level feature map, CUDA out of memory will happen.Is this due to the amount of memory required to compute the Attention matrix? Looking forward...

QAQEthan

关于 Embedded Gaussian问题

4

想问一下，在non_local_embedded_gaussian.py实现文件中，我并没有发现Embedded Gaussian的具体表达式，只是有矩阵乘法+softmax函数...如果是Embedded Gaussian的实现，具体是哪几行代码实现呢？ https://github.com/AlexHex7/Non-local_pytorch/blob/39ad90c91538d34e88865c9fb0ce4a844751346c/lib/non_local_embedded_gaussian.py#L85 https://github.com/AlexHex7/Non-local_pytorch/blob/39ad90c91538d34e88865c9fb0ce4a844751346c/lib/non_local_embedded_gaussian.py#L86

mymuli

AttributeError: 'NoneType' object has no attribute 'fill_'

1

I want to embed this non-local code in my framework, but when I run my code, the following error will occur: > File "/home/Global/models/non_local_embeded_gaussian.py", line 102, in __init__ > super(NONLocalBlock2D,...

Mr-doraemon

理解热力图

1

你好，我也想问一下如何去理解热力图，看了paper还是不太理解~，希望能解答一下~

Createroner

使用点积的热图可视化

1

您好，我用点积版本的non_local对f进行softmax之后， f = torch.matmul(theta_x, phi_x) N = f.size(-1) f_div_C = f / N sft = nn.Softmax(dim=2) f_div_C = sft(f_div_C) 在可视化时还需要对NL_MAP进行normalize吗？如果需要，应该如何normalize呢？ 现在我的热图结果只有一个box，看不到正常热图颜色的变化，这是什么原因呢？

55renchen

implement in other network

2

SO if I want to add nonblock into my own experiment I can only add NONLocalBlock2D() into somewhere of the network ? and I did't need to change other params...

sxzy

how to understand heatmap in data?

1

Hi, why can a heat map be generated?? I don't understant visualization of non-local. Are there any references available？

pantlive

少批量数据有用，大批量数据无效

1

我把non-local 用在reid上面的，在resnet50的前三个layers后面都接了non-local层。用1000个ID进行训练是发现能提高一个点，而且收敛速度比原来快，但是当我用5000个ID训练时发现结果比原来低了一个多点。是不是有哪些参数我需要再调一下？烦请解答哦 PS：我没有发现其他网友提出的W权重为0的情况，我这边W还是挺正常的

guanyonglai

Have tested the nonlocal based on MaskRCNN?

1

https://github.com/facebookresearch/video-nonlocal-net/issues/58 Please refer to this issue.

PkuRainBow