[yolov3] Train your own detector

Training YOLOv3 : Deep Learning based Custom Object Detector

Starting with OpenCV 3.4.2, you can easily use YOLOv3 models in your own OpenCV application.

 

 

Training YOLOv3 : Deep Learning based Custom Object Detector

完整的应用流程。

Code: https://github.com/spmallick/learnopencv/tree/master/YOLOv3-Training-Snowman-Detector

 

YOLOv3 Series 一个系列视频教学。

 

TF训练yolov3，不知道部署有没有问题。

https://github.com/pythonlessons/TensorFlow-2.x-YOLOv3

 

 

 

Why AlexeyAB/darknet

---

前言: 自从Joseph Redmon提出了yolov3后，其darknet仓库已经获得了16k的star，足以说明darknet的流行。该作者最新一次更新也是一年前了，没有继续维护。不过自来自俄国的大神AlexeyAB在不断地更新darknet, 不仅添加了darknet在window下的适配，而且实现了多种SOTA目标检测算法。AlexeyAB也在库中提供了一份详细的建议，从编译、配置、涉及网络到测量指标等，一应俱全。通过阅读和理解AlexeyAB的建议，可以为我们带来很多启发。本文是来自翻译AlexeyAB的darknet中的README，并在翻译的过程中加入我们的一些经验。

 

 

 

 

 

YunYang1994/tensorflow-yolov3

---

ref: https://github.com/YunYang1994/tensorflow-yolov3

 

 

 

 

zzh8829/yolov3-tf2

可能有点参考价值，但以实际为准。

Benchmark (No Training Yet)

Numbers are obtained with rough calculations from detect_video.py

Macbook Pro 13 (2.7GHz i5)

Detection	416x416	320x320	608x608
YoloV3	1000ms	500ms	1546ms
YoloV3-Tiny	100ms	58ms	208ms

Desktop PC (GTX 970)

Detection	416x416	320x320	608x608
YoloV3	74ms	57ms	129ms
YoloV3-Tiny	18ms	15ms	28ms

AWS g3.4xlarge (Tesla M60)

Detection	416x416	320x320	608x608
YoloV3	66ms	50ms	123ms
YoloV3-Tiny	15ms	10ms	24ms

RTX 2070 (credit to @AnaRhisT94)

Detection	416x416
YoloV3 predict_on_batch	29-32ms
YoloV3 predict_on_batch + TensorRT	22-28ms

Darknet version of YoloV3 at 416x416 takes 29ms on Titan X. Considering Titan X has about double the benchmark of Tesla M60, Performance-wise this implementation is pretty comparable.