A Novel Hybrid Integrated Gradient based Self Refined Segment Editing parallel Model for Tracking Entities in Video Frames

This work aims to develop an efficient and interpretable AI system that can accurately detect and monitor entities within restricted spaces, ensuring enhanced security and safety. The proposed model leverages the strength of explainable AI and applies the self-refined segment editing technique, allowing the model to better discern and track persons in complex environments, which enhances the model performance through iterative learning. Combining these two approaches allows the AI system to adapt to various scenarios and refine its tracking capabilities over time. The model is trained and evaluated in the experiment on a custom dataset comprising video frames from multiple restricted areas with posture and gait datasets. The results demonstrate that the hybrid model outperforms traditional tracking methods, exhibiting higher accuracy and interpretability. Additionally, the model's explainable nature provides insights into its decision-making process, enabling better understanding and trust in its output. The proposed hybrid Integrated Gradient with a self-refined segment editing model (IG-SEPM) holds promising potential for applications in security, surveillance, and safety measures in restricted areas. The metrics like mean Average Precision (mAP) (or) Intersection over Union (IoU) are used to measure the segment editing model's accuracy in tracking upon a standard sample. The comparative results show that the proposed model provides better results of about 89.7
^%
Average precision value compared to other existing work. The output from SPEM shows that the model provides a target class predicted probability of 0.3 for α = 0.15. Then the valid masked image is interpreted in parallel using IG and obtains better target class predicted probability with average pixel gradients. The result shows that this proposed work gives better results than the existing system.