piergiaj / pytorch-i3d

piergiaj/pytorch-i3d Issues

How the flo is normalized
Updated 6 months ago
extract feature
Updated 7 months ago2
How to use my own video to extract feature?
Closed a year ago
Why there are no reduce_mean(logits) and 'Predictions' after 'Logits' like Deepmind's version?
Closed a year ago2
How to input different size?
Updated a year ago
How to get per frame embeddings given 1 video
Updated a year ago2
about json
Updated 2 years ago1
Confusion about image normalization
Updated 2 years ago
Root directory structure for custom dataset :
Closed 4 years ago3
What's the in_channels's role?
Updated 2 years ago
number of frames in the input
Updated 2 years ago3
replace_logits function??
Updated 2 years ago
replace_logits function??
Updated 2 years ago2
Could you provide the checkpoints pre-trained on Kinetics-400
Updated 2 years ago1
How to change the frame rate of input for feature extraction?
Updated 3 years ago
How to get the prediction value or accuracy?We get only loss from here but we also need to know accuracy
Updated 3 years ago2
What is the "vid" mean in the code?
Updated 3 years ago1
Why are only a portion of all the frames selected while training?
Updated 3 years ago1
Predict
Updated 3 years ago
Extract Features
Updated 3 years ago4
Extracted Charades features
Updated 3 years ago
Evaluation problem
Closed 5 years ago7
Is there any Pre-trained model for RGB on Kinetics-600?
Updated 3 years ago
extract features
Updated 3 years ago2
any colab version?
Updated 4 years ago
Testing results of I3D model
Updated 4 years ago3
About extract features
Updated 4 years ago
About Kinetics Pretrained
Updated 4 years ago2
Pretraining on Sports-1m
Updated 4 years ago
About the overlapping part.
Updated 4 years ago1
mAP calculation
Updated 4 years ago
Could I share your extracted features for THUMOS 2014?
Updated 4 years ago2
Extending the Inception I3D architecture to TwoStream I3D
Updated 4 years ago
Why is the number of flow frames halved ?
Closed 4 years ago2
Wrong assigning of dropout probability
Updated 4 years ago
Don't you accumulate the validation gradients too while training?
Updated 4 years ago2
Why do you reset the loss every tenth scheduler step, and step the scheduler and zero the gradients every 4th iteration while training on the Charades Dataset?
Updated 4 years ago2
I am using the pre-trained model for feature extraction. I am getting a 4D tensor output. What is mean by each dimensoion
Updated 4 years ago2
extract features of my own videos.
Closed 5 years ago
How can I use pre-trained model to inference?
Updated 5 years ago
Could I share your well-trained models for THUMOS 2014?
Updated 5 years ago
Performance of your ported weights on Kinetics400 validation set (data preprocessing)
Closed 5 years ago2
Overlapping values of end and start
Updated 5 years ago2
I have a question about 64f in the dataloader of the code.
Updated 5 years ago
Final end point
Updated 5 years ago1
File "extract_features_trails.py", line 80, in run b,c,t,h,w = inputs.shape ValueError: not enough values to unpack (expected 5, got 1)
Updated 5 years ago
F.upsample(per_frame_logits)
Updated 5 years ago1
Resuming training with Adam optimiser requires consistent ordering of module's parameters
Updated 5 years ago
Loss function
Updated 5 years ago2
Parameter spatial_squeeze bug
Updated 5 years ago