akhteramin/CaptionOcclusionSeverityEvaluation

Trained YoloV3 using Transfer Learning based on inclusive image dataset(self-annotated) created from live captioned video. The model has been used to identify the salient information region coordinates on the screen followed by the region being overlapped or occluded by captions. A research-based methodological approach has been employed to quantify the level of severity stemmed from this occlusion toward Deaf and Hard of Hearing Viewers.

Language: Python - Size: 467 MB - Last synced at: about 2 years ago - Pushed at: over 3 years ago - Stars: 0 - Forks: 0