We present FoveaBox, an accurate, flexible and completely anchor-free framework for object detection. While almost all state-of-the-art object detectors utilize the predefined anchors to enumerate possible locations, scales and aspect ratios for the search of the objects, their performance and generalization ability are also limited to the design of anchors. Instead, FoveaBox directly learns the object existing possibility and the bounding box coordinates without anchor reference. This is achieved by: (a) predicting category-sensitive semantic maps for the object existing possibility, and (b) producing category-agnostic bounding box for each position that potentially contains an object. The scales of target boxes are naturally associated with feature pyramid representations for each input image. Without bells and whistles, FoveaBox achieves state-ofthe-art single model performance of 42.1 AP on the standard COCO detection benchmark. Specially for the objects with arbitrary aspect ratios, FoveaBox brings in significant improvement compared to the anchor-based detectors. More surprisingly, when it is challenged by the stretched testing images, FoveaBox shows great robustness and generalization ability to the changed distribution of bounding box shapes. The code will be made publicly available.
anchor free設計上挺有意思的，帶來了一些新的思考，正如作者提到的問題一樣：is the anchor boxes scheme the optimal way to guide the search of the objects? 還是需要看哪種prior假設+手調參數更加符合數據的真實分佈，比如有anchor，但設置得很不好，比如牙刷這種ratio相當不常見的物體，anchor相當於提供了一個不準確的初始化，這樣還不如讓網路自己學習角點到中心點的偏差；但如果一些物體本身中心會很難定義是哪一點，比如一個擋住一部分的桌子，讓網路來選擇中心加學習偏差可能又會很confused了，反而不如提供了一個好的anchor來的要好。（小孩子才做選擇，成年人直接兩個都要，都接上然後fuse？233～）
[1] Zhou, Xinyu, et al. "EAST: an efficient and accurate scene text detector." Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. 2017.
[2] Wang, Jiaqi, et al. "Region proposal by guided anchoring." arXiv preprint arXiv:1901.03278 (2019).
