Predicting Video Saliency with Object-to-Motion CNN and Two-layer Convolutional LSTM.
Lai JiangMai XuZulin WangPublished in: CoRR (2017)
Keyphrases
- object motion
- successive frames
- layered representation
- video objects
- space time
- moving objects
- temporal continuity
- motion trajectories
- object detection and tracking
- video sequences
- image sequences
- temporal filtering
- key frames
- video scene
- video frames
- object based video
- video analysis
- spatial and temporal
- visual saliency
- video data
- surveillance videos
- camera motion
- target object
- motion analysis
- camera movement
- cellular neural networks
- motion estimation
- global motion
- video streams
- visual data
- visual input
- dynamic scenes
- video surveillance
- stationary camera
- video content
- motion cues
- depth ordering
- image motion
- object tracking
- d objects
- motion features
- moving camera
- recurrent neural networks
- human motion
- articulated objects
- motion model
- visual attention
- attention mechanism
- restricted boltzmann machine
- motion parameters
- saliency map
- optical flow
- input image
- motion segmentation
- object segmentation
- mean shift
- video shots
- video summarization