小目标检测算法调研

1、Scale Match for Tiny Person Detection

论文链接:https://arxiv.org/abs/1912.10664

代码链接:https://github.com/ucas-vg/TinyBenchmark

专为远距离和大背景下的微小目标设计的检测器

2、RetinaNet升级版:TinaFace

论文链接:https://arxiv.org/abs/2011.13183

代码链接:https://github.com/Media-Smart/vedadet/tree/main/configs/trainval/tinaface

论文解读:https://mp.weixin.qq.com/s/AApYj1djbgz-i-JK_4JHxA

3、Stitcher: Feedback-driven Data Provider for Object Detection

论文链接:https://arxiv.org/abs/2004.12432

代码链接:https://github.com/yukang2017/Stitcher

论文解读:https://mp.weixin.qq.com/s/GS8GhBnFA9l_Q78NcridgQ

4、Finding Tiny Faces in the Wild with Generative Adversarial Network

论文链接:https://openaccess.thecvf.com/content_cvpr_2018/papers/Bai_Finding_Tiny_Faces_CVPR_2018_paper.pdf

论文解读:https://zhuanlan.zhihu.com/p/58990053

5、Augmentation for small object detection

论文链接:https://arxiv.org/abs/1902.07296

代码链接:https://github.com/zzl-pointcloud/Data_Augmentation_Zoo_for_Object_Detection https://github.com/gmayday1997/SmallObjectAugmentation

论文解读:https://mp.weixin.qq.com/s/mNi8qWG9ptIMN77FYV4-Kg

6、其他

https://mp.weixin.qq.com/s/qiGvAW9GH7cEb7LXkuPi8Q

https://zhuanlan.zhihu.com/p/121666693

https://mp.weixin.qq.com/s/NHgV1yUEx1e-C1v7bZwXAg

centernet改进:

1、PPDet:Reducing Label Noise in Anchor-Free Object Detection

论文链接:https://arxiv.org/abs/2008.01167

代码链接:https://github.com/nerminsamet/ppdet

论文解读:https://aijishu.com/a/1060000000131327

2、CPNDet:Corner Proposal Network for Anchor-free,Two-stage Object Detection

论文链接:https://arxiv.org/abs/2007.13816

代码链接:https://github.com/Duankaiwen/CPNDet

论文解读:https://mp.weixin.qq.com/s/941GgrThu61OB09j0xtAHw

3、Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection

论文链接:www.arxiv.org/pdf/2011.1288

代码链接:www.github.com/implus/GFocalV2

论文解读:https://mp.weixin.qq.com/s/H3LuCuqKCUNFldzqiPWQXg

4、DeepMark++: CenterNet-based Clothing Detection

评论 1
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值