An Angular-Temporal Interaction Network for Light Field Object Tracking in Low-Light Scenes
Researchers have developed a novel method for tracking objects in low-light, four-dimensional light field scenes. This approach utilizes a new representation called an epipolar-plane structure image (ESI) to enhance visual details and reduce data redundancy. The proposed angular-temporal interaction network (ATINet) learns from geometric and temporal cues within the light field, and can be optimized through self-supervised learning. A large-scale dataset for light field object tracking in low-light conditions has also been introduced, demonstrating ATINet's state-of-the-art performance. AI
IMPACT Introduces a new method for object tracking in challenging visual conditions, potentially improving autonomous systems and surveillance.