The multi-stream network uses two different streams of information (motion and appearance) for each of two different spatial croppings (full-frame and person-centric) to analyze short chunks of video. One network (CNN-T) computes features on pixel trajectories (motion), while the other (CNN) computes features on RGB channels (appearance).
-
Notifications
You must be signed in to change notification settings - Fork 0
OmerElshrief/Fine-Grained-Action-Detection-Recognition
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
About
A Multi Stream Bi-Directional Recurrent Neural Network for Fine-Grained Action-Detection
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published