Information theory based pruning for CNN compression and its application to image classification and action recognition.
Hai-Hong PhanNgoc-Son VuPublished in: AVSS (2019)
Keyphrases
- information theory
- action recognition
- image classification
- bag of words
- information theoretic
- bag of features
- human detection
- action classification
- statistical learning
- computer vision
- activity recognition
- feature extraction
- spatial temporal
- visual words
- human actions
- conditional entropy
- image features
- image representation
- body parts
- bag of visual words
- visual features
- image compression
- human activities
- recognizing human actions
- class specific
- recognition of human actions
- view invariant
- action detection
- action recognition in videos
- recognizing actions
- spatio temporal
- human pose
- sparse representation