BSUV-Net: A Fully-Convolutional Neural Network for Background Subtraction of Unseen Videos.
Mustafa Ozan TezcanPrakash IshwarJanusz KonradPublished in: WACV (2020)
Keyphrases
- background subtraction
- convolutional neural network
- video surveillance
- moving camera
- static camera
- dynamic scenes
- face detection
- foreground background segmentation
- object detection
- dynamic background
- moving objects
- temporal coherence
- surveillance system
- foreground objects
- image sequences
- motion detection
- background model
- video sequences
- gaussian mixture model
- foreground detection
- human detection
- moving object detection
- shadow detection
- post processing
- background modeling
- appearance model
- background image
- detection of moving objects
- foreground segmentation
- surveillance videos
- machine learning
- traffic monitoring
- video frames
- gait recognition
- foreground regions
- video data
- training examples
- foreground and background
- pixel level
- motion estimation
- face recognition
- neural network
- three dimensional
- detection method
- computer vision
- human actions