Radar vs Camera: Ensuring Reliable Object Detection

Object detection is crucial in robotics, autonomous driving, and surveillance. The reliability of camera-based system is compromised by their inability to function well in unfavorable environmental conditions like fog, snow, or low light. Radar-Based object detection performs better on the other hand in any given condition, Radar-Based detection(RAMP-CNN), performs better with an ap of 82.92% on CRUW dataset but the model performance was incosistent hence we implemented our approach which resulted in ap of 83.52% which performed consistently in any condition, whereas RT-DETR attains 74.8% ap on the CRUW dataset and 45.2% accuracy on the DAWN dataset indicating to be a reliable model in camera based object detection.