RCNN-→SPP Net-→ Fast RCNN-→ Faster RCNN-→ YOLO-→ SSD
思路是：a,生成候选框 b，CNN提取特征 c,分类网络 d,回归，位置精修(refine)
RCNN:
论文:
https://arxiv.org/pdf/1311.2524.pdf
源码:
https://github.com/rbgirshick/rcnn
一些解读：
https://blog.csdn.net/shenxiaolu1984/article/details/51066975
https://blog.csdn.net/hjimce/article/details/50187029
https://blog.csdn.net/tsq292978891/article/details/78722813
SPP Net:
论文:
https://arxiv.org/pdf/1406.4729.pdf
源码:
caffe spp layer:https://blog.csdn.net/u013010889/article/details/53928363
一些解读:
https://zhuanlan.zhihu.com/p/27485018
Fast RCNN
论文:
https://arxiv.org/pdf/1504.08083.pdf
源码:
https://github.com/rbgirshick/fast-rcnn
一些解读:
https://blog.csdn.net/shenxiaolu1984/article/details/51036677
Faster RCNN
论文:
https://arxiv.org/pdf/1506.01497.pdf
源码:
https://github.com/rbgirshick/py-faster-rcnn
一些解读:
https://blog.csdn.net/shenxiaolu1984/article/details/51152614
https://blog.csdn.net/u013010889/article/details/78574879
https://zhuanlan.zhihu.com/p/31426458
YOLO/YOLO.V2
论文:
https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Redmon_You_Only_Look_CVPR_2016_paper.pdf
https://arxiv.org/pdf/1612.08242.pdf
源码:
https://github.com/pjreddie/darknet
一些解读:
https://blog.csdn.net/ben_ben_niao/article/details/52014285
https://blog.csdn.net/u014380165/article/details/72616238
https://blog.csdn.net/jesse_mx/article/details/53925356
SSD
论文:
https://arxiv.org/abs/1512.02325
源码:
https://github.com/weiliu89/caffe/tree/ssd
一些解读:
https://www.cnblogs.com/fariver/p/7347197.html
整个系列总结:
https://blog.csdn.net/linolzhang/article/details/54344350
https://wenku.baidu.com/view/cb977f29f68a6529647d27284b73f242336c31df.html