Object tracking 相关链接汇总(持续更新。。。)

希望对你有用,拿走请点赞。
最全会议期刊的日期: http://conferences.visionbib.com/Iris-Conferences.html
一、框架API
1. pytorch:https://pytorch.org/docs/0.4.0/nn.html?highlight=upsample#torch.nn.functional.upsample
模型地址:https://github.com/pytorch/vision/tree/master/torchvision/models
官方文档:https://pytorch.org/docs/master/torchvision/models.html
通过一个PyTorch项目理解迁移学习:https://www.toutiao.com/a6665489219101655563/?tt_from=mobile_qq&utm_campaign=client_share×tamp=1552008243&app=news_article&utm_source=mobile_qq&iid=37687128300&utm_medium=toutiao_ios&group_id=6665489219101655563
FINETUNING TORCHVISION模型:https://pytorch.org/tutorials/beginner/finetuning_torchvision_models_tutorial.html

二、数据集
全网最大机器学习数据集:
界面优美,方便查找(英文):https://www.datasetlist.com/
中文网址:https://deeplearning4j.org/cn/opendata
单个类别或某个数据集:
youtubb:https://github.com/mbuckler/youtube-bb
基准测试包含在可见光谱内外录制的视频和图像:http://vcipl-okstate.org/pbvs/bench/?tdsourcetag=s_pcqq_aiomsg
OTB:http://cvlab.hanyang.ac.kr/tracker_benchmark/datasets.html
VOT:http://www.votchallenge.net/challenges.html
LaSOT:https://cis.temple.edu/lasot/download.html
new:https://www.epfl.ch/labs/cvlab

  • VID
  • YOUTUBEBB (New link for cropped data, BaiduYun, extract code: 6dd5. NOTE: Data in old link is not correct. Please use cropped data in this new link.)
  • DET
  • COCO
    跟踪方面数据汇总:https://github.com/huanglianghua/siamfc-pytorch/issues/7

三、网络基础操作
1.对深度可分离卷积、分组卷积、空洞卷积的通俗理解(上篇):
https://www.toutiao.com/a6680689513867510286/
对转置卷积、可变形卷积、通道加权卷积的通俗理解(下篇):https://www.toutiao.com/a6680690165775598092/?tt_from=mobile_qq&utm_campaign=client_share×tamp=1555474446&app=news_article&utm_source=mobile_qq&utm_medium=toutiao_ios&req_id=20190417121406010016037041309E3D8&group_id=6680690165775598092
2.OctConv卷积:http://url.cn/5lczRgr
论文地址:https://export.arxiv.org/pdf/1904.05049
第三方复现结果:https://github.com/terrychenism/OctaveConv
3.用自注意力增强卷积:https://url.cn/5ATtyIy
4.33个神经网络技巧:https://url.cn/5ZRO6K2
5.卷积神经网络感受野计算指南:https://mp.weixin.qq.com/s?__biz=Mzg5ODAzMTkyMg==&mid=2247485740&idx=1&sn=ac25bfb0f56e92339eb6d2a016ee9e61&chksm=c0698171f71e0867400593ae2bda78d55e2b65416b77a92e798dbcdea77ae7c456462d87d55f&mpshare=1&scene=23&srcid=#rd
6.目标检测之非极大值抑制(NMS)各种变体:http://url.cn/5wgYIYN
7.一文读懂卷积神经网络中的1x1卷积核:https://mp.weixin.qq.com/s?__biz=MzI0NDUwNzYzMg==&mid=2247484903&idx=1&sn=f013845186776e13b019c9686008ccd2&chksm=e95df378de2a7a6e6451e1938821e4024f6363479239c5690d7c7e9ed2258f8cf88c351f4019&mpshare=1&scene=23&srcid=#rd
8.一文让你看懂转置卷积(反卷积,分数卷积):https://mp.weixin.qq.com/s?__biz=Mzg5ODAzMTkyMg==&mid=2247485790&idx=1&sn=48aea04491702e6fe81fd688e3838b1d&chksm=c0698103f71e0815ff18d4ab1e55e61c0b134bd7fe02464d572123a0f19850d75801832b36be&mpshare=1&scene=23&srcid=#rd
9.注意力模型深度综述:https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247488715&idx=3&sn=1373d850e6e25b5d598b9ce249c4ed38&chksm=ec1ff932db6870249b44cde137e76da9c7b2c42a1799d24c364390f97cd815c52466cd9883dc&mpshare=1&scene=23&srcid=05074nnJqGwRiBNTqnXUfdPD#rd
10.Kaggle实战目标检测奇淫技巧合集:https://mp.weixin.qq.com/s?__biz=MzIwMTE1NjQxMQ==&mid=2247487104&idx=1&sn=a41a6e37be1e169b316d67765d2d9eae&chksm=96f37cd4a184f5c2abb99b5e050642df5e82f172beb19a06c562b91b610f9ed624eb46a2ec81&mpshare=1&scene=23&srcid=0531UHIbLt2e4tWmDWwPrUsE#rd
11.PyTorch 学习笔记(七):PyTorch的十个优化器
(https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247488887&idx=3&sn=f164278eb89d643785f3ab552332818d&chksm=ec1ff88edb68719889036e0118b6bcd0b14a05d5d0709c86df164f205dd277f4fb8429f8dea3&mpshare=1&scene=23&srcid=0521ndCKUBmgxWuLZSSagKdX#rd
12.PyTorch 学习笔记(六):PyTorch的十七个损失函数
(https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247488804&idx=3&sn=e7b30fe16529084a132cdc724e13a5b3&chksm=ec1ff8dddb6871cb783db29a4f8fe2d4bd10ef3da5c125e2b6c9cae7f4cd74568e9670853da6&mpshare=1&scene=23&srcid=05154sktz6FqtxhT2JUpmmjp#rd)
13.深度学习模型-13 迁移学习(Transfer Learning)技术概述
(https://mp.weixin.qq.com/s?__biz=MzIxNDgzNDg3NQ==&mid=2247483947&idx=1&sn=a05a27b35e18b99d3a85c344bbd45e4f&chksm=97a0c9ffa0d740e9eea5005dcc1ae762fa4ecc138ded15816b8f022fb354d0658589cbf9fbef&mpshare=1&scene=23&srcid=0516iQT5nh24YUQYz1ldhq0V#rd
14.PyTorch 学习笔记(三):transforms的二十二个方法
(https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247488626&idx=2&sn=a0e370f879fb32168200dd9988294650&chksm=ec1ff98bdb68709d1ea1a251ab14ea32a72e3e992e5759510edb5dc52ac58a7bb669a729807f&mpshare=1&scene=23&srcid=0430DBNxRgQtAJMg5FPZNxWj#rd

四、相关算法

1.基于siamese network的跟踪算法
http://bbs.cvmart.net/articles/326/ji-yu-luan-sheng-wang-luo-de-gen-zong-suan-fa-hui-zong?from=groupmessage
2.视觉多目标跟踪算法综述
http://url.cn/5gey1IV
3.多目标跟踪 近年论文及开源代码汇总:
https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247488799&idx=1&sn=816a6be8d1abb41e01a6595ed02ed996&chksm=ec1ff8e6db6871f02ac8496916c7d5684efff1081526badebb0899b94095599a153e48680561&mpshare=1&scene=23&srcid=0513LwXGeBBTD5Vjs5AX6OSs#rd

五、相关方向综述
1.目标检测二十年技术综述:
https://mp.weixin.qq.com/s?__biz=MzI5MDUyMDIxNA==&mid=2247488864&idx=1&sn=4d727568d8343288ac76da2409360633&chksm=ec1ff899db68718fadbce2338f6f39a2d31a8fe6b36ede53bc765e2b00362e1c78d8c98bd409&mpshare=1&scene=1&srcid=&key=fcab2858d34cc48f7f8e590b649e5239b3e3589e61c0517bdf75ca9a3beaf244417003dd9fcc6b1fb6285aa9a569d4db5958068c78b0cbcac1922528c23d66d0048d921740787b1e1eaba201e8a0e09d&ascene=1&uin=MjIyMDMyOTgzOA%3D%3D&devicetype=Windows+10&version=62060739&lang=zh_CN&pass_ticket=2L80j4O9b%2BIrFw16q5GU95P0V97yTfVb7oXhetohmQMcc2v1udqD7leipfha5gMQ

六、paper必备
1.一键搞定SCI的参考文献:https://url.cn/578c1Jx

你可能感兴趣的:(跟踪算法链接)