Improved Appearance Model for Handling Occlusion in Vehicle Tracking

Vehicle tracking in computer vision has been an active area of research in recent years. However, tracking a vehicle through occlusion has been a challenging problem due to the loss of visual information. This paper introduces a new approach aimed at enhancing the performance of the current appearance model, specifically GOTURN, which experiences difficulties tracking vehicles during occlusion. The proposed method utilizes a combination of appearance-based tracking and motion-based tracking to improve tracking accuracy even when the vehicle is partially or fully occluded. The proposed method utilizes an appearance-based tracking algorithm (GOTURN) for vehicle tracking until occlusion occurs. Subsequently, a motion-based tracking algorithm (particle filter combined with correlation filter) is employed for tracking the vehicle during occlusion. The approach is assessed on a dataset obtained from various sources, and a comparison is made between the proposed model and other current state-of-the-art methods. The experimental results demonstrate that the proposed approach achieves superior performance in terms of tracking accuracy, especially during occlusion.