Semantic2Graph: Graph-based Multi-modal Feature for Action Segmentation in Videos.
Jun-Bin ZhangPei-Hsuan TsaiMeng-Hsun TsaiPublished in: CoRR (2022)
Keyphrases
- multi modal
- video search
- human actions
- semantic concepts
- image segmentation
- multi modality
- segmentation method
- audio visual
- level set
- video sequences
- segmentation algorithm
- action recognition
- high dimensional
- image features
- computer vision
- medical images
- image annotation
- video analysis
- fully automatic
- medical imaging
- single modality
- fusing multiple
- cross modal
- information theoretic
- video clips
- region growing
- key frames
- video frames
- feature vectors
- multiscale
- high level
- multimedia