Object Detection
A computer-vision task that identifies object classes and estimates their spatial locations, commonly as bounding boxes.
Computer-Vision Context
Object detection predicts both object classes and spatial locations, usually as bounding boxes. In latency-sensitive systems, preprocessing, model inference, non-maximum suppression, and postprocessing should be measured as one pipeline rather than reporting model time alone.
Deployment Boundary
Detection is not image classification because localization is part of the output. Dataset shift, confidence calibration, resolution, and postprocessing thresholds can materially change field performance.