weleen / agrl.pytorch

[TIP2020] Adaptive Graph Representation Learning for Video Person Re-identification

Hi, This is Yiming Wu

agrl.pytorch's Issues

Question about some part of the architecture

I read your 'AGRL' paper many times but I don't understand some parts of it, could you please help me to understand it better

First of all, what do the 'global branch' do and why you use it?
Secondly, after conv5, there is no image, we have some features instead, how did you apply pyramid pooling and match the joints(extracted joints by AlphaPose) in the regions? because in the feature maps the location of joints are not the same as original images.
Third, what is 'N' in the input? If it is the number of regions in the input, there is no region before pyramid pooling I don't understand.

splits_prid2011.json文件怎么获得的啊？

你好！ 
请问下，prid2011中的splits_prid2011.json文件从哪里下载的啊？我从你提供的链接上下不到，可以提供一下文件吗？

data

Can you share the data sets you use ?The link you gave is no longer good.

Where are the pretained Weights?

ValueError: ./MARS/bbox_test/0002/0002C1T0012F001.jpg is not acceptable

Hi~thanks for your code first. I used trained model and pose.json you offered, occur this problem:
Traceback (most recent call last):
File "train_vidreid_xent_htri.py", line 604, in
main()
File "train_vidreid_xent_htri.py", line 331, in main
distmat = test(model, queryloader, galleryloader, args.pool, use_gpu, return_distmat=True)
File "train_vidreid_xent_htri.py", line 504, in test
for batch_idx, (imgs, pids, camids, adj) in enumerate(queryloader):
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 435, in next
data = self._next_data()
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 1085, in _next_data
return self._process_data(data)
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 1111, in _process_data
data.reraise()
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/_utils.py", line 428, in reraise
raise self.exc_type(msg)
ValueError: Caught ValueError in DataLoader worker process 0.
Original Traceback (most recent call last):
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/utils/data/_utils/worker.py", line 198, in _worker_loop
data = fetcher.fetch(index)
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/project/anaconda3/envs/torch171/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py", line 44, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/project/vedioreid/AGRL-master/torchreid/dataset_loader.py", line 208, in getitem
num_scale=self.num_scale, pyramid_part=self.pyramid_part)
File "/project/vedioreid/AGRL-master/torchreid/dataset_loader.py", line 259, in generate_graph
raise ValueError('{} is not acceptable'.format(path))
ValueError: /scratch/vedioreiddata/MARS/bbox_test/0002/0002C1T0012F001.jpg is not acceptable

How to get the pose.json?

I tried to train the model using the dataset ilids-vid, but received the error "No such file or directory: 'data\ilids-vid\pose.json'". It seems that the file "pose.json" is not in the dataset downloaded from the source. How can I get the pose.json of those datasets?

Cross Domain

Thanks for this great work! Can you please share Cross-Domain results of your work?

代码中的超参和论文中略有不同，使用代码中的超参很难训到论文精度，请问有什么训练技巧吗

使用代码中的参数在duke-video上达到了论文精度，在mars上map低于论文四个点，论文中的实验是以那个参数为准？求教，谢谢

self.poses = json.load(f)

对您的这篇文章很感兴趣，但是有一个地方一直有些疑惑，想请教一下您 self.poses = json.load(f)中的bodies中的joints是什么意思啊