【Glossary】
UVOS: Unsupervised Video Segmentation
DVAP (Dynamic Visual Attention Prediction): Dynamic visual attention prediction
AGOS (Attention-Guided Object Segmentation): attention for guiding the cutting
Visual Attention Prediction: predict visual attention, attention mechanisms by people to predict. Early attention model is based on biometric estimate (biologically-inspired features) and cognitive principles (cognitive theories about visual attention), it has recently been recommended and widely used model of attention based on the depth of learning. But a lot of attention model has not been used on UVOS, the article author desires to establish a closer link between the two
(For a detailed explanation of the mechanism of attention here given link)
【Summary】
advantage:
1. model training supervised segmentation is not used, but the use of dynamic positioning data initial training attention training model
2. A detailed understanding of the multi-source learning prospects
3. supplementary explanation of biological stress and attention from the perspective of assessment