Unsupervised Learning of Video Representations via Dense Trajectory Clustering.
Pavel TokmakovMartial HebertCordelia SchmidPublished in: ECCV Workshops (2) (2020)
Keyphrases
- unsupervised learning
- unsupervised manner
- data clustering
- supervised learning
- video sequences
- self organizing maps
- minimum message length
- deep belief networks
- multimedia
- video content
- semi supervised
- dimensionality reduction
- video frames
- video data
- cluster analysis
- deep learning
- object recognition
- point trajectories
- expectation maximization
- video streams
- real time
- data sets
- model selection
- text classification
- space time
- k means
- video retrieval
- video clips
- arbitrary shape
- feature extraction
- clustering algorithm
- machine learning
- data mining
- clustering method
- multimedia data
- key frames
- active learning
- image segmentation