SSD论文阅读(Wei Liu——【ECCV2016】SSD Single Shot MultiBox Detector)
2017-02-24 21:21
573 查看
本文转载自:
http://www.cnblogs.com/lillylin/p/6207292.html
作者及相关链接
文章的选择原因
方法概括
方法细节
相关背景补充
实验结果
与相关文章的对比
总结
intro: ECCV 2016 Oral
arxiv: http://arxiv.org/abs/1512.02325
paper: http://www.cs.unc.edu/~wliu/papers/ssd.pdf
slides: http://www.cs.unc.edu/%7Ewliu/papers/ssd_eccv2016_slide.pdf
github: https://github.com/weiliu89/caffe/tree/ssd
video: http://weibo.com/p/2304447a2326da963254c963c97fb05dd3a973
github(MXNet): https://github.com/zhreshold/mxnet-ssd
github: https://github.com/zhreshold/mxnet-ssd.cpp
github(Keras): https://github.com/rykov8/ssd_keras
性能好,single stage
测试时,输入一张图像到SSD中,网络输出一个下图最右边的tensor(多维矩阵),对该矩阵进行非极大值抑制(NMS)就能得到每个目标的位置和label信息
Figure2的最右图的1th-20th Channel表示类别,每一个Channel上的map对应原图,last 4 channel的每一个map分别对应x,y,w,h的偏移量。最后4个通道可以确定一个box的位置信息,前20个通道确定类别信息。
accurate as Faster R-CNN)
The
core of SSD is predicting category scores and box offsets for a fixed set of default
bounding boxes using small convolutional filters applied to multiple
feature maps from different layers
Experimental evidence: high
accuracy, high speed, simple end-to-end training (single shot)
in bounding box locations
Using separate predictors (filters) for different aspect ratio detections
Using multiple layers for prediction at different scales (apply these
filters to multiple feature maps to perform detection at multiple stages)
http://www.cnblogs.com/lillylin/p/6207292.html
SSD论文阅读(Wei Liu——【ECCV2016】SSD Single Shot MultiBox Detector)
目录
作者及相关链接文章的选择原因
方法概括
方法细节
相关背景补充
实验结果
与相关文章的对比
总结
作者
intro: ECCV 2016 Oral
arxiv: http://arxiv.org/abs/1512.02325
paper: http://www.cs.unc.edu/~wliu/papers/ssd.pdf
slides: http://www.cs.unc.edu/%7Ewliu/papers/ssd_eccv2016_slide.pdf
github: https://github.com/weiliu89/caffe/tree/ssd
video: http://weibo.com/p/2304447a2326da963254c963c97fb05dd3a973
github(MXNet): https://github.com/zhreshold/mxnet-ssd
github: https://github.com/zhreshold/mxnet-ssd.cpp
github(Keras): https://github.com/rykov8/ssd_keras
文章的选择原因
性能好,single stage
方法概括
文章的方法介绍
SSD主要用来解决目标检测的问题(定位+分类),即输入一张待测图像,输出多个box的位置信息和类别信息测试时,输入一张图像到SSD中,网络输出一个下图最右边的tensor(多维矩阵),对该矩阵进行非极大值抑制(NMS)就能得到每个目标的位置和label信息
Figure2的最右图的1th-20th Channel表示类别,每一个Channel上的map对应原图,last 4 channel的每一个map分别对应x,y,w,h的偏移量。最后4个通道可以确定一个box的位置信息,前20个通道确定类别信息。
方法的pipeline和关键点
方法细节
模型结构
多尺度特征图
用来预测的卷积滤波器
defaul box
groundTruth的标定,损失函数
default box和尺度的选择
SSD的训练——Hard negative mining
SSD的训练——数据扩增
相关背景补充
Atrous算法(hole算法)
FPS/SPF, Jaccard overlap
二类分类/检测常用的评价标准 (recall, precision, f-measure, accuracy, error, PR曲线和ROC曲线,AP,AUC)
ImageNet多类分类的评价标准
ImageNet单目标检测的评价标准
ImageNet(多)目标检测的评价标准
实验结果
PASCAL VOC2007 test detection结果
使用数据扩增、多尺度default box、atrous算法的对比效果
SSD512在某类Ianimals)上的检测性能可视化
SSD对于目标大小的敏感性实验
SSD使用的feature map的个数对结果的影响
示例结果
时间和速度
与相关文章的对比
原始R-CNN方法的变形
Faster R-CNN和SSD对比
YOLO和SSD对比
总结
文章贡献
SSD, a single-shot detector for multiple categories (faster than YOLO,accurate as Faster R-CNN)
The
core of SSD is predicting category scores and box offsets for a fixed set of default
bounding boxes using small convolutional filters applied to multiple
feature maps from different layers
Experimental evidence: high
accuracy, high speed, simple end-to-end training (single shot)
SSD对于其他方法的改进的关键点
Using a small convolutional filter to predict object categories and offsetsin bounding box locations
Using separate predictors (filters) for different aspect ratio detections
Using multiple layers for prediction at different scales (apply these
filters to multiple feature maps to perform detection at multiple stages)
相关文章推荐
- SSD论文阅读(Wei Liu——【ECCV2016】SSD Single Shot MultiBox Detector)
- 转:SSD论文阅读(Wei Liu——【ECCV2016】SSD Single Shot MultiBox Detector)
- [置顶] 论文阅读:SSD: Single Shot MultiBox Detector
- 论文阅读:SSD: Single Shot MultiBox Detector
- 论文阅读:SSD: Single Shot MultiBox Detector
- 论文阅读理解 - SSD: Single Shot MultiBox Detector
- 【计算机视觉】目标检测之ECCV2016 - SSD Single Shot MultiBox Detector
- 论文阅读笔记:SSD: Single Shot MultiBox Detector
- 论文阅读:SSD: Single Shot MultiBox Detector
- 论文阅读:SSD: Single Shot MultiBox Detector
- 【论文阅读】SSD: Single Shot MultiBox Detector
- 转载SSD论文阅读:SSD: Single Shot MultiBox Detector
- SSD: Single Shot MultiBox Detector 之再阅读
- SSD:Single Shot MultiBox Detector 论文笔记
- 论文笔记《SSD: Single Shot MultiBox Detector》
- SSD: Single Shot MultiBox Detector 论文笔记
- SSD: Single Shot MultiBox Detector 之再阅读
- Object Detection -- 论文SSD(SSD: Single Shot MultiBox Detector)解读
- 【论文笔记】物体检测系列 SSD: Single Shot MultiBox Detector
- SSD: Single Shot MultiBox Detector 论文笔记