Based on CornerNet, detects each object as a triplet, rather than a pair, of keypoints, CenterNet has made great progress. Do you have a test comparison between CenterNet and CornerNet-Lite? Thank you！

[QQ2737499951]

Based on CornerNet, detects each object as a triplet, rather than a pair, of keypoints, CenterNet has made great progress. Do you have a test comparison between CenterNet and CornerNet-Lite? Thank you！

https://github.com/Duankaiwen/CenterNet

CenterNet: Keypoint Triplets for Object Detection CenterNet：用于对象检测的关键点三元组

CenterNet is an one-stage detector which gets trained from scratch. On the MS-COCO dataset, CenterNet achieves an AP of 47.0%, which surpasses all known one-stage detectors, and even gets very close to the top-performance two-stage detectors.

Description of their paper: “We build our framework upon a representative one-stage keypoint-based detector named CornerNet. Our approach, named CenterNet, detects each object as a triplet, rather than a pair, of keypoints, which improves both precision and recall. Accordingly, we design two customized modules named cascade corner pooling and center pooling, which play the roles of enriching information collected by both top-left and bottom-right corners and providing more recognizable information at the central regions, respectively.”

[xiaozhuka]

我觉得你这是个奇怪的问题，A 论文先发出来，B论文后发出来，B中已经做了详细的对比，你还问A的作者你对比B了吗？

[QQ2737499951]

我觉得你这是个奇怪的问题，A 论文先发出来，B论文后发出来，B中已经做了详细的对比，你还问A的作者你对比B了吗？

Firstly, because the two papers are almost overlapping in time, and they are issued simultaneously, the difference is one day, and CenterNet was submitted on April 17 (v1), the latest revision is April 19, 2019(v3). CornerNet-Lite(Submitted on 18 Apr 2019)

CenterNet: Keypoint Triplets for Object Detection Kaiwen Duan, Song Bai, Lingxi Xie, Honggang Qi, Qingming Huang, Qi Tian (Submitted on 17 Apr 2019 (v1), last revised 19 Apr 2019 (this version, v3))

CornerNet-Lite: Efficient Keypoint Based Object Detection Hei Law, Yun Teng, Olga Russakovsky, Jia Deng (Submitted on 18 Apr 2019)

Secondly, they are based on the original author CornerNet. The core is CornerNet. The original author didn't know this information. I told them the information.

Can't we?

Thirdly, you said that detailed comparison has been made in B. Would you please check the original paper carefully? That's just a comparison with the old version of CornerNet, ok? Is there a comparison between the new CornerNet-Lite?

Fourthly, in the spirit of research, if CornerNet-Lite and CornerNet-Lite+CenterNet were compared, would there be any difference?

Thank you!

第一，因为这两篇论文的时间几乎是重叠的，同时发的，相差1天，而且CenterNet是4月17日提交（v1），最新修订于2019年4月19日。CornerNet-Lite4月18日提交 CenterNet：用于对象检测的关键点三元组 Kaiwen Duan, Song Bai, Lingxi Xie, Honggang Qi, Qingming Huang, Qi Tian （2019年4月17日提交（v1），最新修订于2019年4月19日（本版本，v3））

CornerNet-Lite：基于关键点的高效对象检测 Hei Law，Yun Teng，Olga Russakovsky，Jia Deng （于2019年4月18日提交）

第二，CenterNet和CornerNet-Lite都是基于原始作者CornerNet分别改进而来的。核心都是CornerNet，原始作者可能不知道这个信息，我告知一下不可以吗？

第三，你说B中已经做了详细的对比，麻烦您仔细核对原始论文，那只是对比CornerNet的老版本，ok？新的CornerNet-Lite有对比吗？

第四，本着研究的精神，如果CornerNet-Lite与CornerNet-Lite+CenterNet对比会不会有差异呢？

谢谢！

[xiaozhuka]

@QQ2737499951 谢谢你的分析，我没有留意到你提到的细节信息，抱歉我的回答 还有一个建议给你，既然都开源了，也可以下载到作者的模型，可以自己分析下啊

[QQ2737499951]

谢谢！正在验证，还没有结果

[xcaizewu]

@xiaozhuka 想请问您个问题，作者给出的ConnerNet-lite的安装教程，需要利用conda创建一个python3.7的虚拟环境，那么在conda_packagelist.txt中需要安装的包，是都安装在虚拟环境中，还是有一部分安装到系统里？因为我现在其他的虚拟环境中在跑faster，我担心安装这些包时会对已有虚拟环境中安装的包有影响