Coder Social home page Coder Social logo

rcnn's Introduction

RCNN

Rich feature hierarchies for accurate object detection and semantic segmentation

工程内容

这个程序是基于tensorflow的tflearn库实现部分RCNN功能。

开发环境

windows10 + python3.5 + tensorflow1.2 + tflearn + cv2 + scikit-learn

数据集

采用17flowers据集, 官网下载:http://www.robots.ox.ac.uk/~vgg/data/flowers/17/

程序说明

1、setup.py---初始化路径
2、config.py---配置
3、tools.py---进度条和显示带框图像工具
4、train_alexnet.py---大数据集预训练Alexnet网络,140个epoch左右,bitch_size为64
5、preprocessing_RCNN.py---图像的处理(选择性搜索、数据存取等)
6、selectivesearch.py---选择性搜索源码
7、fine_tune_RCNN.py---小数据集微调Alexnet
8、RCNN_output.py---训练SVM并测试RCNN(测试的时候测试图片选择第7、16类中没有参与训练的,单朵的花效果好,因为训练用的都是单朵的)

文件说明

1、train_list.txt---预训练数据,数据在17flowers文件夹中
2、fine_tune_list.txt---微调数据2flowers文件夹中
3、1.png---直接用选择性搜索的区域划分
selectivesearch_1
4、test/2.png---通过RCNN后的区域划分
RCNN_1

程序问题

1、由于数据集小的原因,在微调时候并没有像论文一样按一个bitch32个正样本,128个负样本输入,感觉正样本过少;
2、还没有懂最后是怎么给区域打分的,所以没有NMS,待续;   3、对选择的区域是直接进行缩放的;
4、由于数据集合论文采用不一样,但是微调和训练SVM时采用的IOU阈值一样,有待调参。

参考

1、论文参考:
https://www.computer.org/csdl/proceedings/cvpr/2014/5118/00/5118a580-abs.html
2、代码参考:
http://www.cnblogs.com/edwardbi/p/5647522.html
https://github.com/edwardbi/DeepLearningModels/tree/master/RCNN    
3、博客参考:
http://blog.csdn.net/u011534057/article/details/51218218
http://blog.csdn.net/u011534057/article/details/51218250        

rcnn's People

Contributors

yangxue0827 avatar

Watchers

James Cloos avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.