RecurrentAttentionConvolutionalNeuralNetwork icon indicating copy to clipboard operation
RecurrentAttentionConvolutionalNeuralNetwork copied to clipboard

Published 20 hours ago • verified publisher icon dillondavis

ask for help

Open 22wei22 opened this issue 6 years ago • 6 comments

Sorry， I want to know can the code run？ how to get dataset，Thanks ModuleNotFoundError: No module named 'torchnet' where is the torchnet

22wei22 avatar Apr 24 '18 07:04 22wei22

This repos a work in progress at the moment. It requires the CUB dataset. torchnet is a python dependency that can be installed using pip It should be runnable in its current state but its not currently working.

dillondavis avatar Apr 24 '18 17:04 dillondavis

what I get is the images of CUBS,where to find such as image_crop_labels2.txt,image_class_labels.txt'

22wei22 avatar Apr 25 '18 12:04 22wei22

You have to generate candidate crops for each image in CUBS and pretrain the attention proposal networks with these candidate crops. I'm trying out multiple ways at the moment because the paper does not say how they efficiently did this process. My approaches to generate these candidates are in utils.py which generates image_crop_labels.txt

dillondavis avatar May 01 '18 16:05 dillondavis

when i run the code of graph_networks, with SummaryWriter(comment='RACNN3') as w: AttributeError: enter thanks

22wei22 avatar May 03 '18 03:05 22wei22

graph_networks.py is experimental at the moment. I tried using it to visualize some of my networks with tensorboardx but there are some bugs.

dillondavis avatar May 07 '18 04:05 dillondavis

I want to know which version of Pytorch you used?My version is pytorch-1.0.0.

xmuwenxiang avatar Dec 19 '18 03:12 xmuwenxiang