<ul><li>Researchers introduce VRU-CIPI framework to predict Vulnerable Road Users' crossing intentions at intersections.</li><li>The framework utilizes a sequential attention-based model with GRU and Transformer self-attention mechanism.</li><li>VRU-CIPI achieves a high accuracy of 96.45% on UCF-VRU dataset, with real-time inference speed of 33 frames per second.</li><li>Integrating with Infrastructure-to-Vehicles communication enhances intersection safety by activating crossing signals and warning connected vehicles.</li></ul>

VRU-CIPI: Crossing Intention Prediction at Intersections for Improving Vulnerable Road Users Safety

Discover more