Object detection is the computer vision task of not only classifying objects in an image but also locating them with bounding boxes. An object detector outputs a set of bounding boxes, each with a class label and usually a confidence score. Early approaches to detection involved sliding window classifiers (like running a CNN or SVM on many crops of the image). Modern detectors are typically of two kinds: two-stage detectors (like Faster R-CNN) which first propose candidate regions (Region Proposal Network) and then classify/refine them; and single-stage detectors (like YOLO, SSD) which directly predict boxes and classes on a dense grid of possible locations. Evaluation often uses metrics like mAP (mean Average Precision at certain IoU thresholds). Challenges in detection include dealing with scale variance (objects can be small or large), viewpoint changes, intra-class variation, and needing to be efficient for potentially many objects per image. Modern detectors can run in real-time enabling applications like autonomous driving (detecting pedestrians, vehicles), video surveillance, and more.