SegTAD: Precise Temporal Action Detection via Semantic Segmentation.
Chen ZhaoMerey RamazanovaMengmeng XuBernard GhanemPublished in: ECCV Workshops (4) (2022)
Keyphrases
- semantic segmentation
- action detection
- superpixels
- conditional random fields
- action recognition
- atomic actions
- scene classification
- object detection
- spatio temporal
- pascal voc
- object categories
- temporal relations
- image sequences
- pixel level
- space time
- image classification
- probabilistic model
- object classes
- computer vision
- image representation
- natural images
- feature space
- multi class
- image features