search

hassony2 / kinetics_i3d_pytorch

Inflated i3d network with inception backbone, weights transfered from tensorflow
MIT License
528 stars 115 forks source link

Arch mod #27

Closed albanie closed 4 years ago

albanie commented 5 years ago

Thanks for sharing the implementation! These are just a few small changes to achieve numerical consistency with the TF version by:

  1. Modifying the padding algorithm to match Tensorflow "SAME" for Unit3Dpy and MaxPool3dTFPadding.
  2. Updating the batch norm hyperparams.

Tested on PyTorch 0.4.1. Running the i3d_pt_demo.py produces:

Top 5 classes and associated probabilities:
[playing cricket]: 9.999968E-01
[playing kickball]: 1.335340E-06
[catching or throwing baseball]: 4.553116E-07
[shooting goal (soccer)]: 3.143419E-07
[catching or throwing softball]: 1.924329E-07
Top 5 classes and associated probabilities:
[playing cricket]: 9.477569E-01
[hurling (sport)]: 4.068211E-02
[playing tennis]: 4.154132E-03
[playing squash or racquetball]: 2.474060E-03
[hitting baseball]: 1.380014E-03
===== Final predictions ====
logits proba class
4.181368e+01 1.000000e+00 playing cricket
2.149398e+01 1.497148e-09 hurling (sport)
2.013410e+01 3.843058e-10 catching or throwing baseball
1.922558e+01 1.549212e-10 catching or throwing softball
1.891534e+01 1.135999e-10 hitting baseball
hassony2 commented 4 years ago

@albanie I stopped closely following this repo a long while ago and it looks like I missed this pull request, although it appears super useful and also looks related to a recently raised issue !

It is said that better late than never :) (and on our time-scales one and a half year is arguably close to never !)

Have a great day !

Best,

Yana