How does YOLO / SSD / RetinaNet differ from two-stage detection?
mediumAnswer
- One-stage detectors predict class and box in a single forward pass over dense anchor / grid locations — no proposal step.
- YOLO: divide image into an S×S grid, each cell predicts B boxes + class probs.
- SSD: multi-scale feature maps with different anchor scales.
- RetinaNet: adds Focal Loss to fix the extreme class imbalance (background vs objects).
- Much faster, real-time capable; historically less accurate but the gap narrowed with recent versions.
Check yourself — multiple choice
- One-stage uses proposals
- Predict class + box directly over dense anchor locations — faster, no proposal stage (YOLO, SSD, RetinaNet)
- Slower than Faster R-CNN
- No anchors used
One-stage: predict class + box over dense anchors → fast, real-time detection.
#detection#computer-vision#losses
Practise Deep Learning
214 interview questions in this topic.
Related questions
- Explain focal loss and why it's used in dense detection.
- How does a two-stage detector (Faster R-CNN) work?
- What is Non-Maximum Suppression (NMS) and its variants?
- What are anchor boxes and their downsides?
- How does Mask R-CNN extend Faster R-CNN for instance segmentation?
- What is a Feature Pyramid Network (FPN)?