What is object detection in the context of ADAS perception?

Object detection identifies and locates objects of interest in sensor data. For cameras, this involves classifying image regions as vehicles, pedestrians, cyclists, traffic signs, etc., and drawing bounding boxes around them. Modern systems use deep learning (CNNs like YOLO, SSD, Faster R-CNN) trained on large datasets. For radar and lidar, detection involves identifying clusters of returns that represent objects and estimating their position, size, and motion. Output includes object class, position (x, y, z), dimensions, velocity, and confidence score. Detection must be fast (10-30 Hz) and reliable, handling occlusion, weather, and lighting variations.