vividle / a2net Goto Github PK

Hi, thanks for your excellent and inspiring work.
The anchor-based part got a promising result on on thumos14 dataset. However, when I tried it on anet dataset, only got a average map of about 13, which is much lower than the result reported on paper.
So I wonder if you adpoted the same network architecture in anet as in thumos14, looking forward to your reply, thanks.

Hi, I read your paper A2Net and it is really amazing work on action localization.

However, when I downloaded the feature on ActivityNet and tried to run it, it has the error that:

'begin_frame' is not a file in the archive.

and I manually looked into the feature file, it seems that for ActivityNet, the file does not have the 'begin_frame' key.

Can you explain that part?
Also, I want to know that the material you provides including 'cuhk_simple_share.json'. It is the classification result for ActivityNet. I want to know where does it come from? Is it come from the anet-cuhk 2016 or anet-cuhk 2017?

Thanks!

I'm having a bit of a problem reproducing my dataset with your code. Multiple GT messages are required under lib/dataset/prepare_data/construct_feature_thumos.py. I didn't find the corresponding place in your code to modify which prevented me from proceeding.

@VividLe Hi，thanks for your sharing, I have download the ActivityNett1.3 feature.And how do you get 'cls_label', 'reg_label'? Could you share me the preprocess script? Thank you very much!!

Hello, thank you very much for sharing. I have some questions about I3D feature extraction.
I extract features through the links provided by you. I feel that the dimension of output features is not well understood. Assuming that the scale of the input I3D is [4,3,256,224,224] and the dimension of the final output is [4,1024,31,1,1], in the code you provide, the feature extracted by I3D should get a value in each frame. if its need to upsample?