Supplementary Appendix for ”Improved Stereo Matching with Constant Highway Networks and Reflective Confidence Learning”
نویسندگان
چکیده
Previous architectures in the literature for computing the matching cost report no benefit from using color information [3, 1]. In our experiments we observed that after deepening our network, the use of the three input channels contributes to the accuracy of the disparity prediction, especially around areas of delicate color differences between the object and its background. An example for this phenomenon is shown in Fig. 1, and the average improvements over the validation sets of KITTI and Middlebury are presented in Tab. 1.
منابع مشابه
Optimizing Disparity Candidates Space in Dense Stereo Matching
In this paper, a new approach for optimizing disparity candidates space is proposed for the solution of dense stereo matching problem. The main objectives of this approachare the reduction of average number of disparity candidates per pixel with low computational cost and high assurance of retaining the correct answer. These can be realized due to the effective use of multiple radial windows, i...
متن کاملLeveraging Stereo Matching with Learning-based Confidence Measures (Supplementary Material)
1. Confidence measures We used a variety of confidence measures in [1–5] for constructing the 22-dimensional feature vector f. Before explaning each of them, we introduce a few notations to explain confidence measures in a simple and clear way. Instead of using C(p, d) as a matching cost, we use ci as the i th minimum matching cost among all disparity hypotheses D for p. di is the corresponding...
متن کاملColor Matching of Blends Prepared From Black and White Fibers by Neural Networks (TECHNICAL NOTE)
The color of the blends of pre-colored fibers depends on the ratio of each fiber in the blends. Some theories have been introduced for color matching of blends of pre-colored fibers. Most however, are restricted in scope and accuracy. Kubelka and Munk presented the most applicable theory, which is still used in industry. In this work, the classical Kubelka-Munk method for color prediction of a ...
متن کاملA CNN Solution for Depth Estimation from Binocular Stereo Imagery
Novel results and experiments are presented on the application of cellular neural networks to binocular stereo vision. A CNN universal machine algorithm is described for depth estimation as a part of a stereo-vision-based guidance system for autonomous vehicles. Being most amenable to revealing stereo correspondence, extraction of vertical edges is performed first. Then their distance from the ...
متن کاملReflective Learning and Teaching: A Review
Introduction: One of the most important characteristic of human being is his ability to learn. Structuralists believe that learning is an active process through which learners explores the principles, meanings and facts by themselves. Learner’s participation in learning process is one of the active learning strategies and reflective learning is considered as an active teaching method which is i...
متن کامل