德克萨斯大学提出：One-stage目标检测最强算法 ExtremeNet
德克萨斯大学提出：One-stage目标检测最强算法 ExtremeNet
前戏
最近出了很多论文，各种SOTA。比如（点击可访问）：
今天头条推送的是目前人脸检测方向的SOTA论文：改进SRN人脸检测算法。本文要介绍的是目前(2019-01-26) one-stage目标检测中最强算法：ExtremeNet。
正文
《Bottom-up Object Detection by Grouping Extreme and Center Points》
arXiv: https://arxiv.org/abs/1901.08043
github: https://github.com/xingyizhou/ExtremeNet
作者团队：UT Austin
注：2019年01月23日刚出炉的paper
Abstract：With the advent of deep learning, object detection drifted from a bottom-up to a top-down recognition problem. State of the art algorithms enumerate a near-exhaustive list of object locations and classify each into: object or not. In this paper, we show that bottom-up approaches still perform competitively. We detect four extreme points (top-most, left-most, bottom-most, right-most) and one center point of objects using a standard keypoint estimation network. We group the five keypoints into a bounding box if they are geometrically aligned. Object detection is then a purely appearance-based keypoint estimation problem, without region classification or implicit feature learning. The proposed method performs on-par with the state-of-the-art region based detection methods, with a bounding box AP of 43.2% on COCO test-dev. In addition, our estimated extreme points directly span a coarse octagonal mask, with a COCO Mask AP of 18.9%, much better than the Mask AP of vanilla bounding boxes. Extreme point guided segmentation further improves this to 34.6% Mask AP.
Illustration of our object detection method
Illustration of our framework
Illustration of our object detection method
基础工作
- Extreme and center points
- Keypoint detection
- CornerNet
- Deep Extreme Cut
创新点
- Center Grouping
- Ghost box suppression
- Edge aggregation
- Extreme Instance Segmentation
实验结果
ExtremeNet有多强，看下面的图示就知道了，在COCO test-dev数据集上，mAP为43.2，在one-stage detector中，排名第一。可惜的是没有给出时间上的对比，论文中只介绍说测试一幅图像，耗时322ms（3.1 FPS）。
State-of-the-art comparison on COCO test-dev