search

Gait3D / Gait3D-Benchmark

This is the code for the paper "Gait Recognition in the Wild with Dense 3D Representations and A Benchmark. (CVPR 2022)", "Gait Recognition in the Wild with Multi-hop Temporal Switch", and "Parsing is All You Need for Accurate Gait Recognition in the Wild".
137 stars 19 forks source link

I have a few questions about Gait3D-Benchmark Datasets #10

Closed jjunnii closed 2 years ago

jjunnii commented 2 years ago

Hi. Im jjun. I read your paper impressively.

We don't currently live in China, so it is difficult to use dataset on baidu disk.

If you don't mind, is there a way to download the dataset to another disk (e.g Google drive)?

JinkaiZheng commented 2 years ago

Hi, @jjunnii. We are so sorry that we don't have any other disk at present. But we will upload it to Onedrive later, and we will inform you at that time, thank you for your understanding!

jjunnii commented 2 years ago

Thank you for your response. I have one more question. I'm going to conduct an experiment with my custom dataset, but it's hard to understand the data structure, so what kind of information do you have on the /lib/modeling/models/smplgait.py. line41 image What data does seql have here? Is the sequence length per identity? Specifically, I wonder what kind of data structure seql has.

JinkaiZheng commented 2 years ago

Hi, seqL is a list of the lengths of each sequence used during the test phase. Because we load all frames at test time, and each sequence length is different.

jjunnii commented 2 years ago

Thank you very much. Lastly, should I set seql to None during the training time? Impressed and the current study under way into your paper. Your advice is likely to be a great help.

JinkaiZheng commented 2 years ago

Yes, seqL can be set to None. Because the training process doesn't use it. We fix the length of each sequence for stable training.

jjunnii commented 2 years ago

Thank you. Your reply was very helpful.