What is Single Shot MultiBox Detector (SSD), and how does it differ from YOLO?
SSD (Single Shot MultiBox Detector) is, like YOLO, a one-stage detector that predicts all boxes in a single pass. Its distinguishing idea is making predictions from several feature maps of decreasing resolution, so different layers naturally handle different object sizes, using a set of default boxes per location. Early YOLO predicted from a single scale; both families have since converged on multi-scale prediction.
SSD and YOLO were the two detectors that made real-time, single-shot detection mainstream, and they share a lot of DNA. The differences are in the details.
- Multi-scale by design: SSD attaches detection heads to several feature maps of different resolutions, so small objects are caught on fine maps and large ones on coarse maps.
- Default boxes: SSD uses per-location default boxes (its version of anchors) with a range of aspect ratios.
- Single pass: like YOLO, it's fast, with no region-proposal stage.
The original YOLO predicted from one coarse grid, which hurt it on small objects; SSD's multi-scale heads handled small objects better at the time. Modern YOLO versions adopted multi-scale necks (FPN/PAN) and anchor-free heads, and have generally overtaken SSD on the speed-accuracy trade-off, which is why new projects almost always reach for a current YOLO rather than SSD.
Related questions
Hands-on tutorials
Stuck on this in a real project?
Book a free call and I’ll give you a straight answer on your specific case.
Book a consultation