[论文笔记]Speed/accuracy trade-offs for modern convolutional object detectors
来源:互联网 发布:new balance跑鞋 知乎 编辑:程序博客网 时间:2024/06/08 06:06
论文链接:https://arxiv.org/pdf/1611.10012.pdf
这篇论文是由谷歌发表的,主要在速度和准确性两方面来权衡目标检测算法的性能,主要比较了三个经典的目标检测算法(Faster-rcnn, SSD, R-FCN),这三种算法分别使用六种特征提取器来实现检测( VGG-16, Resnet-101, Inception v2, Inception v3, Inception Resnet (v2), MobileNet ),使用开源框架Tensorflow。
三种算法原理图
虽然在SSD和R-FCN论文中都讲到无论是在速度上还是在准确度上都要比Faster-rcnn要好,但是在同样的网络结构,训练方法等条件下,该论文证明其实还是Faster-rcnn的准确性要好,但是速度方面却不占优势。看下图
上图可以看出还是Faster-rcnn(圆圈)的mAP比较高,但是速度却是慢了些。
下图展示的是不同算法对不同大小物体(large,medium,small)的检测精度,可以看出大目标的检测精度就是比小目标的检测精度要高,Faster-rcnn的效果也最好(多数情况下)
在特征提取器比较中,Inception Resnet (v2) 结构是效果最好的,看下图
可以看出没有一种算法可以在速度和准确性上都做的很好,本篇文章给出了根据自己的目标选择的依据(看中速度还是看中准确性还是两者兼顾),下面是几张效果图
阅读全文
0 0
- 论文笔记之---Speed and accuracy trade-offs for modern convolutional object detectors
- [论文笔记]Speed/accuracy trade-offs for modern convolutional object detectors
- Speed/accuracy trade-offs for modern convolutional object detectors
- Speed/accuracy trade-offs for modern convolutional object detectors
- Speed/accuracy trade-offs for modern convolutional object detectors
- 论文阅读-《Speed/accuracy trade-offs for modern convolutional object detectors》
- READING NOTE: Speed/accuracy trade-offs for modern convolutional object detectors
- 目标检测“Speed/accuracy trade-offs for modern convolutional object detectors”
- 论文:Speed/accuracy trade-offs for object detectors
- object_detection API论文Speed/accuracy trade-offs for...的详细中文翻译
- DSOD: Learning Deeply Supervised Object Detectors from Scratch 论文笔记
- 论文研读--LocNet: Improving Localization Accuracy for Object Detection
- Region-based Convolutional Networks for Accurate Object Detection and Segmentation----R-CNN论文笔记
- 论文笔记 MSCNN:A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection
- 论文笔记《IS object localization for free? Weakly-supervised learning with convolutional neural network》
- 【论文笔记】Region-based Convolutional Networks for Accurate Object Detection and Segmentation
- 论文笔记(二) fully-convolutional siamese networks for object tracking
- 【论文笔记】(CVPR2016 Oral) Training Region-based Object Detectors with Online Hard Example Mining
- 关于mybatis传参问题
- Oracle 12c Windows安装、介绍及简单使用(图文)
- 数据导出到excel
- Cmake是如何工作的
- capstone训练营2017-12-21
- [论文笔记]Speed/accuracy trade-offs for modern convolutional object detectors
- Okhttp请求封装类
- laravel服务提供者与服务容器的因果关系
- TensorFlow各版本下载地址,强烈推荐
- Effective C++之七:模板与泛型编程
- 三种方法重置 mysql 数据库自增ID
- ConcurrentHashMap理解
- org.springframework.web.servlet.handler.AbstractHandlerMethodMapping$MappingRegistry异常的解决方法
- C++ 库研究笔记——Linux下创建文件夹