Improvements of 3D Object Detection with Part-based Models
-
摘要: 三维目标检测问题是计算机视觉领域的一个基础而重要的问题, 如何解决部分遮挡、类内变化、复杂背景以及视角变化的问题是这类算法的研究重点. 近年来,随着部分遮挡、类内变化等问题的逐步解决,越来越多的研究者针对视角问题 展开研究.本文对三维目标检测问题进行了较为详细的分析,并且主要针对近几年的 热点问题—视角问题展开讨论,介绍并总结了当前该领域的主要算法. 通过对比说明了各种方法的优势与不足.Abstract: 3D Object detection in 3D views is important and fundamental in the computer vision field. Its core issues include coping with the problems of partial occlusion, intra-class variance, cluttered background and multiple views. Most recently, with the development of part-based model, the occlusion and intra-class variance problems have been partly solved. Therefore, researchers have focused their attention mainly on multi-view problem. This paper gives detailed analysis of the 3D object detection problem with emphasis on the recent works related to this problem. At last, we summarize the algorithms introduced and give their comparisons.
-
Key words:
- 3D object /
- object detection /
- part-based /
- geometry model /
- pose estimation /
- multi-view
点击查看大图
计量
- 文章访问数: 2172
- HTML全文浏览量: 61
- PDF下载量: 1424
- 被引次数: 0