Classification says what is in an image. Detection says what and where. Object detection draws bounding boxes around each instance and labels it, which is why it matters for surveillance, driving, and retail analytics.

Two families dominate. Two-stage detectors like Faster R-CNN propose regions first, then classify them. One-stage detectors like YOLO and SSD predict boxes and classes in a single pass. One-stage models run faster; two-stage models often edge them on accuracy.

Common detection tasks

• Pedestrian and vehicle detection
• Face detection
• Defect detection in manufacturing
• Wildlife monitoring
• Retail shelf analysis

Small objects and crowded scenes remain hard. A detector that finds cars on an empty highway may miss them in a traffic jam. Domain-specific data and augmentation help close the gap.