【Deep Learning】Review of Stereo Matching by Training a Convolutional Neural Network to Compare Image
来源:互联网 发布:python中延时函数 编辑:程序博客网 时间:2024/06/10 05:46
Stereo Matching by Training a Convolutional Neural Network to Compare Image
Link: http://arxiv.org/abs/1510.05970
Code: https://github.com/jzbontar/mc-cnn
_____________________________________
1. Summary of thePaper
By presenting a matching cost computation, the paperprovides a similarity measurement extracting depth information from a pairpictures via convolutional neural network. The speed model could give resultswithin a second while the accuracy model lower the error rate on Kitti.
2. MainContributions
1) Asthe paper itself indicated, a description of two architectures (One for speedand the other for accuracy.) based on convolutional neural networks forcomputing the stereo matching cost.
2) Amethod, accompanied by its source code, with the lowest error rate on the KITTI2012, KITTI 2015, and Middlebury stereo data sets.
3) Experimentsanalyzing the importance of data set size, the error rate compared with othermethods, and the trade-off between accuracy and runtime for different settingsof the hyper- parameters.
3. Positive andnegative points
Positive Points:
(i) Itprovides two models that either gave a fastest speed ever or gave the mostaccurate rate ever since.
(ii) Itdemonstrates very detailed dataset augmentation methods, which, not novelthough, are necessary to achieve good results.
Negative Points:
(i) It actuallydidn’t present brand-new idea of improving the stereo matching but indeedprovides us an overall perspective of the whole process.
4. How strong isthe evaluation
Extremelyimpressive. The accurate MC-CNN-acrt model ranks first amony all method on theKITTI 2012. Even the MC-CNN-fst model ranks 5th and the runtime islargely beyond any other methods.
As for KITTIstereo ranking, these two methods totally dominate others as they rank 1stand 2nd lowest error rate. However, ELAS methods is slightly quickerthan MC-CNN-fst.
However, theMC-CNN-fst cannot show up in the Middlebury stereo data set but stillMC-CNN-acrt ranks 1st without doubt.
5. Possibledirection for the future work
Maybe they canpursue the measurement of similarity for a video.
- 【Deep Learning】Review of Stereo Matching by Training a Convolutional Neural Network to Compare Image
- 论文学习——Stereo Matching by Training a Convolutional Neural Network to Compare Image Patches
- Stereo Matching by Training a Convolutional Neural Network to Compare Image Patches
- [ICCV2017]Cascade Residual Learning: A Two-stage Convolutional Neural Network for Stereo Matching
- Computing the Stereo Matching Cost with a Convolutional Neural Network
- A new deep convolutional neural network for fast hyperspectral image classification Review
- 立体匹配之(二):[MC-CNN] 2015CVPR: Stereo Matching by Training a Convolutional Neural Netw
- Learning to Compare Image Patches via Convolutional Neural Networks
- Learning a Deep Convolutional Network for Image Super-Resolution(泛读)
- Learning a Deep Convolutional Network for Image Super-Resolution(泛读)
- Deep Convolutional Neural Network for Image Deconvolution
- Deep-Learning NotePad3 : convolutional neural network
- 【论文笔记】Image Classification with Deep Convolutional Neural Network
- Learning a Deep Convolutional Network for Image Super-Resolution—Chao Dong_ECCV2014
- ISSCC 2017论文导读 Session 14 Deep Learning Processors,A 2.9TOPS/W Deep Convolutional Neural Network
- 论文笔记 Ensemble of Deep Convolutional Neural Networks for Learning to Detect Retinal Vessels in Fundus
- 基于2-channel network的图片相似度判别_2015Learning to Compare Image Patches via Convolutional Neural Networks
- Unsupervised Learning of Stereo Matching
- 【杭电】[2023]求平均成绩
- 使用 jQuery Deferred 和 Promise 创建响应式应用程序
- 一月学习总结
- JavaScript 开发进阶:理解 JavaScript 作用域和作用域链
- 【杭电】[2030]汉字统计
- 【Deep Learning】Review of Stereo Matching by Training a Convolutional Neural Network to Compare Image
- 【Deep Learning】Review of Designing Deep Networks for Surface Normal Estimation
- Nginx安装与配置
- Java类与文件
- CGAL1_1 Three Points and One Segment
- 安装完ubuntu系统后要做的事(以ubuntu15.10为例)
- 【Leetcode】Rectangle Area && Classic Prob: Overlap Rectangle
- 七小时 Theano 入门(Day 2)
- 【Leetcode】Search a 2D Matrix II