Abstract:
Driver assistance and intelligent autonomous vehicles will benefit from automatic human activity recognition and prediction. In this article, we present a novel single image driver action recognition algorithm inspired by human perception that selectively focuses on parts of images to acquire information at specific task-specific locations. Human activity is a combination of pose and semantic contextual cues, unlike other approaches. We model this by considering body joint configuration and their pairwise relation with objects to capture structural information. Our semantically rich body-pose and body-object interaction representation is highly discriminative even with a basic linear SVM classifier. We propose a Multi-stream Deep Fusion Network (MDFN) to combine CNN features and high-level semantics. Our experiments show that the proposed approach significantly improves drivers’ action recognition accuracy on two exacting datasets.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
