需要细细研究的论文

 

 

1、CenterNet(无anchor 检测:bbox keypoint)
知乎:https://zhuanlan.zhihu.com/p/62836519
论文:https://arxiv.org/pdf/1904.07850.pdf
github:https://github.com/xingyizhou/CenterNet
同时还有FoveaBox:
博文:https://mp.weixin.qq.com/s/DoN-vha1H-2lHhbFOaVS8w
paper path:https://arxiv.org/pdf/1904.03797v1.pdf

2、morphNet(加速网络)
博文:https://mbd.baidu.com/newspage/data/landingsuper?context={"nid"%3A"news_9720362154603662894"}&n_type=0&p_from=1
github:https://github.com/google-research/morph-net

3、octConv(新卷积方式)
博文:https://mbd.baidu.com/newspage/data/landingsuper?context={"nid"%3A"news_9720362154603662894"}&n_type=0&p_from=1
论文:https://export.arxiv.org/pdf/1904.05049
github:https://github.com/terrychenism/OctaveConv

4、YOLACT: Real-time Instance Segmentation(realtime分割)
博文:https://blog.csdn.net/jiaoyangwm/article/details/89176767
github:https://github.com/dbolya/yolact

5、TimeCycle(无监督跟踪和光流新方法)
博文:https://mp.weixin.qq.com/s/g0tgdRXtR-i1Hf6v6-r1zw
论文:Learning Correspondence from the Cycle-Consistency of Time
github:https://github.com/xiaolonw/TimeCycle
Slides (google drive): drive.google.com/file/d
Slides (百度网盘): https://pan.baidu.com/share/init?surl=prNthUokiqRPELs8J4O-vQ

6、2DASL(3D人脸重建与密集对齐,比PRNet更优秀)
博文:https://mp.weixin.qq.com/s/h8tk28IEuqB7g1e-vsyN5g
论文:Joint 3D Face Reconstruction and Dense Face Alignment from A Single Image with 2D-Assisted Self-Supervised Learning
https://arxiv.org/pdf/1903.09359.pdf
github:https://github.com/XgTu/2DASL

7、高级语义特征提取(检测精度更高,主要针对行人检测)
论文:High-level Semantic Feature Detection:A New Perspective for Pedestrian Detection
paper path:https://arxiv.org/abs/1904.02948v1
github:https://github.com/liuwei16/CSP

8、实时-视频流检测:
博文:https://mp.weixin.qq.com/s/Sq3dBuU8aY3Ug9NBZMc5lA
论文:https://arxiv.org/abs/1903.10172
github:https://github.com/tensorflow/models/tree/master/research/lstm_object_detection

9、ThunderNet(旷视、更快检测)
博文:https://mp.weixin.qq.com/s/F0ykoKv027ycinsAZZjbWQ
论文:ThunderNet: Towards Real-time Generic Object Detection

10、ZQCNN vs libfacedetection(快速人脸检测)
博文:https://mp.weixin.qq.com/s/DkvQ86tgU4DarsmY_qTLdg
github:https://github.com/zuoqing1988/ZQCNN-MTCNN-vs-libfacedetection
github2:https://github.com/ShiqiYu/libfacedetection

人脸对比可以使用faiss库进行微秒级别的搜索

人脸识别:argface、mtcnn

评论
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