本文共 3666 字，大约阅读时间需要 12 分钟。

YOLO-Summary

YOLO源码：

• https://github.com/pjreddie/darknet
• https://github.com/AlexeyAB/darknet
  非常推荐AlexeyAB的darknet改进版
  论文：
• https://pjreddie.com/media/files/papers/YOLOv3.pdf

YOLOv3复现代码合集涵盖 5 种常用深度学习框架：

TensorFlow

Project	Infernece	Train	star
	√	√	1837
	√	√	795
	√	x	666
	√	√	272

PyTorch

Project	Infernece	Train	star
	√	√	2955
	√	√	2686
	√	x	2291
	√	x	1489
	√	√	1471
	√	√	442
	√	√	258

Keras

Project	Infernece	Train	Star
	√	√	4680
	√	x	505
	√	√	410

Caffe

Project	Infernece	Train	Star
	√	√	569
	√	x	273
	√	√	163

MXNet

Project	Infernece	Train	Star
	√	√	3187

参考：

• https://zhuanlan.zhihu.com/p/50170492
• https://github.com/amusi/YOLO-Reproduce-Summary/blob/master/README.md

一、yolo框架的解读：

• https://zhuanlan.zhihu.com/p/32525231

二、500问里目标检测解决的问题和yolo解读

• https://github.com/scutan90/DeepLearning-500-questions

三、基于YOLO的项目

3.1使用YOLOv3训练、使用Mask-RCNN训练、理解ResNet、模型部署、人脸识别、文本分类等：

• https://github.com/StevenLei2017/AI_projects

3.2基于yolo3 与crnn 实现中文自然场景文字检测及识别

• https://github.com/chineseocr/chineseocr

3.3 YOLOv3 in PyTorch > ONNX > CoreML > iOS

• https://github.com/ultralytics/yolov3

3.4YoloV3/tiny-YoloV3+RaspberryPi3/Ubuntu LaptopPC+NCS/NCS2+USB Camera+Python+OpenVINO

• https://github.com/PINTO0309/OpenVINO-YoloV3

四、YOLO模型压缩：

4.1、剪枝：

• https://github.com/zbyuan/pruning_yolov3
• https://github.com/coldlarry/YOLOv3-complete-pruning
• https://github.com/Lam1360/YOLOv3-model-pruning
• https://github.com/tanluren/yolov3-channel-and-layer-pruning

五、YOLO系列

5.1 Enriching Variety of Layer-wise Learning Information by Gradient Combination

Model	Size	mAP@0.5	BFLOPs
EfficientNet_b0-PRN	416x416	45.5	3.730
EfficientNet_b0-PRN	320x320	41.0	2.208

• https://github.com/WongKinYiu/PartialResidualNetworks

5.2 Gaussian YOLOv3: An Accurate and Fast Object Detector Using Localization Uncertainty for Autonomous Driving

• https://github.com/jwchoi384/Gaussian_YOLOv3

5.3 YOLO Nano: a Highly Compact You Only Look Once Convolutional Neural Network for Object Detection

Model	model Size	mAP(voc 2007)	computational cost(ops)
Tiny YOLOv2[13]	60.5MB	57.1%	6.97B
Tiny YOLOv3[14]	33.4MB	58.4%	5.52B
YOLO Nano	4.0MB	69.1%	4.57B

• https://arxiv.org/pdf/1910.01271.pdf
• https://github.com/liux0614/yolo_nano

5.4YOLO-LITE: A Real-Time Object Detection Algorithm Optimized for Non-GPU Computers

DataSet	mAP	FPS
PASCAL VOC	33.57	21
COCO	12.26	21

• https://arxiv.org/abs/1811.05588v1
• https://github.com/reu2018dl/yolo-lite
• https://mp.weixin.qq.com/s/xNaXPwI1mQsJ2Y7TT07u3g

5.5 SlimYOLOv3: Narrower, Faster and Better for Real-Time UAV Applications

• https://arxiv.org/ftp/arxiv/papers/1907/1907.11093.pdf
• https://github.com/PengyiZhang/SlimYOLOv3
• https://mp.weixin.qq.com/s/fDOskKqG-fsJmhT0-tdtTg

5.6 Strongeryolo-pytorch - Pytorch implementation of Stronger-Yolo with channel-pruning

Performance on VOC2007 Test(mAP) after pruning

Model	Backbone	MAP	Flops(G)	Params(M)
strongerv3	Mobilev2	79.6	4.33	6.775
strongerv3-sparsed	Mobilev2	77.4	4.33	6.775
strongerv3-Pruned(30% pruned)	Mobilev2	77.1	3.14	3.36
strongerv2	Darknet53	80.2	49.8	61.6
strongerv2-sparsed	Darknet53	78.1	49.8	61.6
strongerv2-Pruned(20% pruned)	Darknet53	76.8	49.8	45.2

• https://github.com/wlguan/Stronger-yolo-pytorch

5.7 Learning Spatial Fusion for Single-Shot Object Detection

System	test-dev mAP	Time (V100)	Time (2080ti)
	33.0	20ms	24ms
YOLOv3 608+	37.0	20ms	24ms
YOLOv3 608(ours baseline)	38.8	20ms	24ms
YOLOv3 608+ ASFF	40.6	22ms	28ms
YOLOv3 608+ ASFF*	42.4	22ms	29ms
YOLOv3 800+ ASFF*	43.9	34ms	40ms

• https://arxiv.org/pdf/1911.09516.pdf
• https://github.com/ruinmessi/ASFF

5.8 Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression

• https://arxiv.org/pdf/1911.08287.pdf
• https://github.com/Zzh-tju/DIoU-darknet
• https://mp.weixin.qq.com/s/St5WevfcVt4RubJsY-ZEHw

转载地址：http://rezlf.baihongyu.com/