A Real-Time Background Replacement Method Based on Estimated Depth for AR Applications.
Yoshihiro TsubokiTomoya KawakamiSatoru MatsumotoTomoki YoshihisaYuuichi TeranishiPublished in: COMPSAC (2022)
Keyphrases
- real time
- clustering method
- classification method
- experimental evaluation
- data sets
- high precision
- augmented reality
- computational cost
- segmentation algorithm
- detecting moving objects
- camera tracking
- segmentation method
- support vector machine svm
- vision system
- input data
- cost function
- significant improvement
- pairwise
- preprocessing
- multiscale
- face recognition