Rough Sets Theory with Deep Learning for Tracking in Natural Interaction with Deaf

Sign languages commonly serve as an alternative or complementary mode of human communication Tracking is one the most fundamental problems in computer vision, and use a long list applications such sign recognition. Despite great advances recent years, tracking remains challenging due to many factors including occlusion, scale variation, etc. The mistake detecting head left hand instead right overlapping are, modes like this, uncertainty area over deaf news video frames; we proposed two methods: first, using particle filter second idea rough set theory granular information with deep neural network. We method for Combination Rough Set Deep Neural Network used Hand/Head Video Signal DeafNews. develop system Deaf News. increase accuracy skin segmentation signal. Using network, extracted inherent relationships available frame pixels generalized achieved features tracking. tested on 33 News 100 different words 1927 files then recall, MOTA MOTP values are obtained.