请选择 进入手机版 | 继续访问电脑版

易学智能

 找回密码
 立即注册

QQ登录

只需一步,快速开始

查看: 1735|回复: 3

德克萨斯大学提出:One-stage目标检测最强算法 ExtremeNet

[复制链接]

81

主题

162

帖子

493

积分

推广大使

积分
493
发表于 2019-1-26 15:34:08 | 显示全部楼层 |阅读模式
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》

1.jpg


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.

2.jpg

3.jpg
Illustration of our object detection method

4.jpg

Illustration of our framework

5.jpg
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)。

6.jpg
State-of-the-art comparison on COCO test-dev



回复

使用道具 举报

665

主题

1234

帖子

6561

积分

xdtech

Rank: 5Rank: 5

积分
6561
发表于 2019-2-20 23:30:30 | 显示全部楼层
越做越强了
回复

使用道具 举报

665

主题

1234

帖子

6561

积分

xdtech

Rank: 5Rank: 5

积分
6561
发表于 2019-2-20 23:30:35 | 显示全部楼层
是否已经开源了呢?
回复

使用道具 举报

665

主题

1234

帖子

6561

积分

xdtech

Rank: 5Rank: 5

积分
6561
发表于 2019-2-20 23:30:44 | 显示全部楼层
没有代码,没法用上啊
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

QQ|Archiver|手机版|小黑屋|易学智能

GMT+8, 2024-3-29 00:18 , Processed in 0.015518 second(s), 23 queries .

Powered by Discuz! X3.4

© 2001-2017 Comsenz Inc.

快速回复 返回顶部 返回列表