This post summarize the Object Detection model: SSD - Single Shot Multibox Detector (Liu et al, 2016).
Alongside with region proposal detector like: R-CNN, Fast R-CNN, Faster R-CNN, SSD eliminates the region proposal stage so that it’s fast and also faster than YOLO.
The SSD approach is based on a feed-forward convolutional network that produces a fixed-size collection of bounding boxes and scores for the presence of object class instances in those boxes, followed by a non-maximum suppression step to produce the final detections. It has some key features:
- Multi-scale feature maps for detection.
- Convolutional predictors for detection.
- Default boxes and aspect ratio.
