目录
解释了yolov3的forward和anchor box相关内容
github上7k星的pytorch教程(对不起,现在已经8k星了):
分割线 -------------------------------------
服装识别开源实例(看到新的会再补充,这部分最近有可能会关注一下):
模型转化工具(支持TF,Keras,Pytorch,MXNet多种主流框架):
CVPR2019 | 29篇目标检测相关论文汇总(部分含源码)
Pytorch代码调试利器,自动print Tensor信息:
交通标志数据集Mapillary Traffic Sign Dataset:
LFFD 一个轻量级的检测器(人脸/人头/行人/车牌/车辆):
高斯YOLOv3:使用定位不确定性进行自动驾驶的精确快速目标检测器:
CrowdHuman+Double Anchor:强强联合,推动密集行人检测技术落地:
vedaseg:A semantic segmentation toolbox in pytorch:
yolo系列汇总:
关于yolov1讲的比较清楚的博客:
https://blog.csdn.net/hrsstudy/article/details/70305791
关于yolov1原理解释的幻灯片:
关于anchor box讲的比较清楚的博客:
https://blog.csdn.net/zijin0802034/article/details/77685438
- 注意作者文中说的:线性变换,宽度和高度必需相等,指的是proposal和ground truth的宽和高要近似相等,不能相差太多才是线性变换。
关于yolov2讲的比较清楚的博客:
https://blog.csdn.net/zijin0802034/article/details/77097894
- 里面提到了对偏移量tx 和ty做logistic的原因
关于yolov3的部分解读:
https://xmfbit.github.io/2018/04/01/paper-yolov3/
- 里面提到了梯度如何计算的问题
关于batch normalization的理解:
https://www.cnblogs.com/guoyaohua/p/8724433.html
各类归一化方法的总结及代码:
https://blog.csdn.net/liuxiao214/article/details/81037416
YJango的卷积神经网络介绍:
https://zhuanlan.zhihu.com/p/27642620
目标检测SSD讲解:
https://zhuanlan.zhihu.com/p/33544892
关于AP PR曲线计算的理解(内附代码,生肉,英文):
https://github.com/rafaelpadilla/Object-Detection-Metrics#create-the-ground-truth-files
解释了yolov3的forward和anchor box相关内容
解释了源码yolo_layer.c line195~237,待验证):
https://blog.csdn.net/weixin_41015185/article/details/84189515
关于数据增强的奇淫技巧:
https://www.zhihu.com/question/35339639
一篇超级全的数据增强博客:
https://blog.csdn.net/daniaokuye/article/details/78535879
yolov3 pytorch教程:
- 英文原版生肉(大概能解决你一半以上关于Yolov3的困惑)
https://blog.paperspace.com/how-to-implement-a-yolo-object-detector-in-pytorch/
- 知乎翻译中文
https://zhuanlan.zhihu.com/p/36899263
- 对应github(快去给他小星星吧)
https://github.com/ayooshkathuria/YOLO_v3_tutorial_from_scratch
标注工具汇总:
https://blog.csdn.net/chaipp0607/article/details/79036312
github上7k星的pytorch教程(对不起,现在已经8k星了):
https://github.com/yunjey/pytorch-tutorial
github上2w+星的tensorflow教程:
https://github.com/aymericdamien/TensorFlow-Examples
- 不喜欢英文的可以看这个,pytorch 0.4中文文档:
经典论文原文及其中文对照翻译:
- Alexnet VGG Resnet GoogLenet BN-GoogLenet Inception-v3 SENet
- YOLO SSD YOLO9000 Deformable-ConvNets Faster R-CNN R-FCN FPN
- CRNN CTPN
https://github.com/jiajunhua/SnailTyan-deep-learning-papers-translation
分割线 -------------------------------------
数据增强代码:
手写数字识别博客:
https://blog.csdn.net/qq_33000225/article/details/73123880
keras搭建cnn实现手写数字识别:
https://www.jianshu.com/p/b22e708a3f37
关于残差网络的解析:
https://blog.csdn.net/rogerchen1983/article/details/79353972
pytorch官方教程翻译:
HyperLPR:深度学习高性能车牌识别
https://github.com/zeusees/HyperLPR
服装开源数据集(6w测试集1w数据集):
https://github.com/zalandoresearch/fashion-mnist/tree/master/data/fashion
服装识别开源实例(看到新的会再补充,这部分最近有可能会关注一下):
https://github.com/KaiJin1995/fashionAI2018/tree/master
损失函数各类交叉熵讲解(从信息量和熵来推导,浅显易懂)
https://www.jianshu.com/p/47172eb86b39
https://blog.csdn.net/red_stone1/article/details/80735068
交通灯数据集:
博世小交通灯数据集(用于深度学习的小型交通灯数据集):
https://hci.iwr.uni-heidelberg.de/node/6132
LaRa交通灯识别(巴黎交通灯数据集):
http://www.lara.prd.fr/benchmarks/trafficlightsrecognition
WPI数据集(交通灯/行人/车道检测数据集):
http://computing.wpi.edu/dataset.html
交通信号灯/交通指示牌数据集:
https://github.com/ytzhao/Robotics/wiki/TS-and-TL-Dataset
模型可视化工具(支持多种主流框架):
https://github.com/lutzroeder/Netron
CVPR2019论文:
https://github.com/extreme-assistant/cvpr2019
SSD讲解:
https://blog.csdn.net/qianqing13579/article/details/82106664
一文搞懂faster-rcnn:
https://zhuanlan.zhihu.com/p/31426458
模型转化工具(支持TF,Keras,Pytorch,MXNet多种主流框架):
https://github.com/Microsoft/MMdnn
AI学习资料整理:
CVPR2019 | 29篇目标检测相关论文汇总(部分含源码)
边缘检测论文集合(含代码):
基于yolov3和crnn的ocr识别:
https://github.com/chineseocr/chineseocr
Python-Opencv入门:
http://www.1zlab.com/article/opencv-face-detection/
Pytorch代码调试利器,自动print Tensor信息:
https://github.com/zasdfgbnm/TorchSnooper
coco2014数据集下载地址:
国外:https://blog.csdn.net/u013249853/article/details/84924808
国内:https://blog.csdn.net/weixin_43599336/article/details/86539042
车道线检测论文及代码汇总:
https://github.com/amusi/awesome-lane-detection
模型所需计算力统计工具pytorch-OpCounter:
https://github.com/Lyken17/pytorch-OpCounter
CNN模型解读合集:
https://zhuanlan.zhihu.com/p/65564209
李宏毅机器学习在线阅读笔记:
https://datawhalechina.github.io/leeml-notes
微软亚洲研究院开源项目:
https://mp.weixin.qq.com/s/EDDeWQdfEeh8tISMpPOsCQ
交通标志数据集Mapillary Traffic Sign Dataset:
https://www.mapillary.com/dataset/trafficsign
Pytorch中文教程:
Torchcv 基于Pytorch的开源CV框架:
https://github.com/donnyyou/torchcv
LFFD 一个轻量级的检测器(人脸/人头/行人/车牌/车辆):
原版:https://github.com/YonghaoHe/A-Light-and-Fast-Face-Detector-for-Edge-Devices
Pytorch版:https://github.com/becauseofAI/lffd-pytorch
高斯YOLOv3:使用定位不确定性进行自动驾驶的精确快速目标检测器:
https://github.com/jwchoi384/Gaussian_YOLOv3
PAN 一个快速的文本检测算法:
https://github.com/WenmuZhou/PAN.pytorch
1MB轻量级人脸带关键点检测算法:
https://github.com/biubug6/Face-Detector-1MB-with-landmark?tdsourcetag=s_pctim_aiomsg
目标检测数据集转换工具:
https://github.com/spytensor/prepare_detection_dataset
YOLO系列资料整合:
https://github.com/Bubble-water/YOLO-Summary
一个RCNN系列讲解超级棒的博客:
http://www.telesens.co/2018/03/11/object-detection-and-classification-using-r-cnns/
Pytorch官方出品的一本深度学习的书:
https://zhuanlan.zhihu.com/p/93116220
新鲜开源ASFF,一个贼吊的目标检测器:
https://github.com/ruinmessi/ASFF
Pytorch2Tensorrt的python API:
https://github.com/NVIDIA-AI-IOT/torch2trt
ICCV2019辅助驾驶资料汇总:
https://mp.weixin.qq.com/s/PFPGX4dDtj_1A4nvtBNFRA
YOLO的各种变种:
ASFF: Learning Spatial Fusion for Single-Shot Object Detection
https://github.com/ruinmessi/ASFF
Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression
https://github.com/Zzh-tju/DIoU-darknet
Gaussian YOLOv3: An Accurate and Fast Object Detector Using Localization Uncertainty for Autonomous Driving
https://github.com/jwchoi384/Gaussian_YOLOv3
PRN: Enriching Variety of Layer-wise Learning Information by Gradient Combination
https://github.com/WongKinYiu/PartialResidualNetworks
一个更优秀的backbone:Cross Stage Partial Networks
https://github.com/WongKinYiu/CrossStagePartialNetworks
彩蛋:Centernet也可以用Darknet实现
https://github.com/CaoWGG/CenterNet-darknet
Pytorch训练加速123--如何给cpu数据读取加速:
针对CPU到GPU的张量迁移工具:
https://zhuanlan.zhihu.com/p/82666665
给训练踩踩油门 —— Pytorch 加速数据读取:
https://zhuanlan.zhihu.com/p/80695364
在深度学习中喂饱gpu:
https://zhuanlan.zhihu.com/p/77633542
基于GAN的黄种人人脸生成器:
https://github.com/a312863063/seeprettyface-generator-yellow
Anchor Free论文合集:
https://github.com/VCBE123/AnchorFreeDetection
CrowdHuman+Double Anchor:强强联合,推动密集行人检测技术落地:
https://zhuanlan.zhihu.com/p/95253096
首个使用 NAS 设计的 GCN,达到动作识别SOTA:
https://github.com/xiaoiker/GCN-NAS
NVIDIA AI TOT团队的实时人体姿态估计:
https://github.com/NVIDIA-AI-IOT/trt_pose
基于CenterNet魔改的姿态估计CenterPose:
https://github.com/tensorboy/centerpose
YOLACT++ 开源实时实例分割网络:
https://github.com/dbolya/yolact
UCF-Crime Dataset:
https://webpages.uncc.edu/cchen62/dataset.html
零售商品检测的基准:密集目标检测的强大baseline:
https://github.com/ParallelDots/generic-sku-detection-benchmark
vedaseg:A semantic segmentation toolbox in pytorch:
https://zhuanlan.zhihu.com/p/98561946
何凯明新作,视频理解代码开源:
https://github.com/facebookresearch/SlowFast?tdsourcetag=s_pctim_aiomsg
一个暴力事件检测的数据集:
https://github.com/ZHEQIUSHUI/Violence-Recognition-Dataset
身份证要素提取:
https://github.com/Mingtzge/2019-CCF-BDCI-OCR-MCZJ-OCR-IdentificationIDElement
口罩检测:
https://aijishu.com/a/1060000000091759?from=timeline&isappinstalled=0
轻量级中文OCR识别,模型仅17M:
https://github.com/ouyanghuiyu/chineseocr_lite
表情识别数据库RaFD:
http://www.socsci.ru.nl:8180/RaFD2/RaFD?p=main