The main benefit of YOLO is its speed. While it can achieve very good results, it is now outperformed by more complex networks. Faster Region with Convolutional Neural Networks (Faster R-CNN) is considered state of the art at the time of writing. It is also quite fast, reaching 4-5 FPS on a modern GPU. In this section, we will explore its architecture.
The Faster R-CNN architecture was engineered over several years of research. More precisely, it was built incrementally from two architectures—R-CNN and Fast R-CNN. In this section, we will focus on the latest architecture, Faster R-CNN:
- Faster R-CNN: towards real-time object detection with region proposal networks (2015), Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun
This paper draws a lot of knowledge from the two previous designs. Therefore, some of the...
