leaderj1001 / action-localization Goto Github PK

Hi,

I am having trouble understanding this part of the code.
if inds.numel() > 0: out_inds = torch.arange(inds[-1].item(), inds[-1].item() + inds.numel() // 1) cls_inside_weights[inds] = 1 cls_inside_weights[out_inds] = 1

https://github.com/leaderj1001/Action-Localization/blob/master/action-transformer/nets/proposal_target_layer.py#L67

As I understand this is to set the weights for the classification loss.

However, I do not understand the value of out_inds. Why is it an array of values from inds[-1].item(), inds[-1].item() + inds.numel()? This seems very arbitrary, I cannot understand the logic behind this

Is it ok to set the classification loss weights all to 1? such as
cls_inside_weights[:] = 1
?

Thank you!

Thank you a lot for your contribution. However, I find several pickle files used in 'Action-Localization-master/action-transformer/data/ava.py' confusing, such as 'ori_action_list.pkl' and 'fps_dict.pkl'. Can you show me how you created these files?

(1) Do I need to use "make_tubelet" files in "action-baseline" and "action-transformer" ?
(2) Where are the files used in "action-baseline" and "action-transformer" projects obtained? For example "ori_action_list.pkl", "fps_dict.pkl", and "Main/train.txt"?

when i set batch_size>1, it is wrong.

Hi,

Thanks very much for making your code available! I noticed a lot of of external files are also required which aren't made available, such as:

ava_train_v2.2.pkl
transformer_ava_train_v2.2.pkl
fps_dict.pkl
annotation/train.txt
ori_action_list.pkl

Will it be possible to share what these files contain and how we may obtain/recreate them! Thanks very much!

what's the project used for？

• ava_file_names_test_v2.1.txt
• ava_file_names_trainval_v2.1.txt