zhengyuzhao / adi-red-scene Goto Github PK

当加载 whole_wideresnet18_places365_python36.pth.tar model的时候，总是出现
ModuleNotFoundError: No module named 'wideresnet' 错误，不知道怎么回事

Do you think there is any innovation in scene classification?
Is it convenient to add a wechat to communicate? My wechat is vv19970403

Hello I want to know how to train this network with my own data set

Hello，I am reading this paper recently. I don't know how to upsample the Dis-Map into the image size.

demo.py use the model is places-CNN，and the secene result is :
probs, idx = h_x.sort(0, True)
but this model is pretrained model from placee365 (wideresnet18_places365.pth.tar)， where your model ?

The demo.py can not get the result.

Hello my friend！I am reproducing your entire experiment, but there are some problems when running your code, I cannot open the follow URL
os.system ('wget https://raw.githubusercontent.com/csailvision/places365/master/wideresnet.py'+' -P ./examples')
to download the wideresnet model, Could you please upload wideresnet.py to this folder？Thank you!

Hi,I am reading this paper recently,in your paper ,There is such a sentence in your paper that we use two different patch sizes (1/4 and 1/16 of the image size).but the implementation part of the code is described like this top=np.int32(aratio)-112 left=np.int32(bratio)-112 left=np.maximum(left,0) right=left+224 ,...can you offer some suggestions? thank you !

When I run the file "dis_map_generation", I adjusted the batch_size to 64, but there will be an error "size mismatch", Can batch_size only be set to 256? and why?