Question about how to extract feature

antoine77340 / Mixture-of-Embedding-Experts

Mixture-of-Embeddings-Experts

Apache License 2.0

118 stars 15 forks source link

Question about how to extract feature #12

Open sharontaozi opened 5 years ago

sharontaozi commented 5 years ago

In the paper,the appearance features are extracted using ResNet-152,and the motion features are computed using a Kinetics pre-trained I3D flow network. I really want to know how this part is practiced. If you like, can you share the implemented code or ideas for this part with me?

akonstantinou commented 5 years ago

I am also interested on the extraction of the features. Are you going to give us the implemented code?

antoine77340 commented 4 years ago

Hi all, Sorry for the delayed answer,

you can find: The code for the Resnet-152 extraction code here: https://github.com/antoine77340/video_feature_extractor VGG audioset here: https://github.com/tensorflow/models/tree/master/research/audioset and for I3D flow, I unfortunately do not plan to release the full pipeline from Optical flow to feature extraction as this is quite cumbersome to release. I just followed the original github instructions here: https://github.com/deepmind/kinetics-i3d

DeepaliVerma commented 3 years ago

Sir, can you please explain me how the facial features are extracted from the MSR-VTT dataset.

antoine77340 commented 3 years ago

Hi,

I've used the face recognition feature from http://dlib.net/