akash2907 / pi-consistency-activity-detection Goto Github PK

First of all, thanks for sharing this great work!

I have tried to reproduce the performance of UCF-101-24 using your code, but it failed.

(100 epoch, [email protected] : 90.02 vs 95.1 (in the paper) / [email protected] : 84.62 vs 89.1 (in the paper)

In addition, the current version of the code is different from what you described in your README.md, so I'm having trouble reproducing the experiment.

It would be grateful if you could tell me how to run the experiments of UCF-101-24 based on the current version of the code.

Thanks for your awesome work!
Cloud you provide the test_annots.pkl for test ?

Do you have environment configuration？

Hi,
Nice work!
But I can't find the IoULoss in the utils/losses.py. Could you please update it?
Thank you!

Cloud you provide the train_annots_20_labeled.pkl and train_annots_10_labeled.pkl ? The result of our modification based on the train_annots_20_labeled.pkl file has an error

Hi,
Nice work!
But I don't know -e /home/ak119590/gait/training_codes/exp_1_basic_3d_train/transformers_train/timesformer in requirements.

How do I get it？

Thank you for your excellent work on this paper！
Could you please provide the dataset annotation format of the following annotation files ?

-testing_annots.pkl
-train_annots_20_labeled_random.pkl
-train_annots_80_unlabeled_random.pkl

When I run the main_ucf101.py code, the following error declaration occurs.How can this be solved?
line 146, in train_model_interface
total_cons_loss = total_seg_cons_loss
UnboundLocalError: local variable 'total_seg_cons_loss' referenced before assignment

At line 133 of main.py, the variable 'loss_wt_grad' is not define.I can't find a definition about this variable.

total_seg_cons_loss_2 = (wt_ramp * loss_wt_grad) + ((1 - wt_ramp) * loss_wt_simple_l2)