快速去阴影--Fast Shadow Detection from a Single Image Using a Patched Convolutional Neural Network
2017-09-30 08:56
891 查看
Fast Shadow Detection from a Single Image Using a Patched Convolutional Neural Network
https://arxiv.org/abs/1709.09283
本文主要解决快速去阴影问题,这里使用的策略是 SVM+CNN
A. Computing Shadow Prior
首先使用 mean shift 算法对输入图像进行过分割,得到 segment, 对每个 segment 我们提取其 color and texture features 信息输入 SVM 得到 shadow prior 就是 the log likelihood output of this trained classifier
B. Training Patched-CNN with the Shadow Prior
基于文献【17】,我们使用一个 Patched-CNN 来 predict shadow,其输入是 shadow prior (P) 和对应的 RGB image,输出是 shadow probability map of the patch
C. Edge Refinement of Super-Pixel Labels
上一步主要是 region, 阴影的边界 的 prediction 比较差,所以我们这里 process the edge pixels between the regions by patched-CNN once again,We only process those pixels that are on edges between the segments
Experiments
速度还是比较慢啊!
https://arxiv.org/abs/1709.09283
本文主要解决快速去阴影问题,这里使用的策略是 SVM+CNN
A. Computing Shadow Prior
首先使用 mean shift 算法对输入图像进行过分割,得到 segment, 对每个 segment 我们提取其 color and texture features 信息输入 SVM 得到 shadow prior 就是 the log likelihood output of this trained classifier
B. Training Patched-CNN with the Shadow Prior
基于文献【17】,我们使用一个 Patched-CNN 来 predict shadow,其输入是 shadow prior (P) 和对应的 RGB image,输出是 shadow probability map of the patch
C. Edge Refinement of Super-Pixel Labels
上一步主要是 region, 阴影的边界 的 prediction 比较差,所以我们这里 process the edge pixels between the regions by patched-CNN once again,We only process those pixels that are on edges between the segments
Experiments
速度还是比较慢啊!
相关文章推荐
- Fast Neural Cell Detection Using Light-Weight SSD Neural Network译文
- 目标检测--A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection
- Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural
- 人群计数:Single-Image Crowd Counting via Multi-Column Convolutional Neural Network(CVPR2016)
- Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
- Learning Depth from Single Monocular Images Using Deep Convolutional Neural Fields
- Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
- 人群计数--Single-Image Crowd Counting via Multi-Column Convolutional Neural Network
- Image Style Transfer Using Convolutional Neural Network
- 论文笔记:Single-Image Crowd Counting via Multi-Column Convolutional Neural Network
- [论文解读] MSCNN: A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection
- 多尺度R-CNN论文笔记(5): A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection
- 快速多尺度人脸检测--Multi-Scale Fully Convolutional Network for Fast Face Detection
- Fast convolutional neural network training using selective data sampling: Application to hemorrhage
- 论文笔记-深度估计(1)Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
- Fast convolutional neural network training using selective data sampling: Application to hemorrhage
- [论文解读] Vehicle Detection from 3D Lidar Using Fully Convolutional Network
- 论文阅读《Edge Detection Using Convolutional Neural Network》
- 【论文笔记】Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
- 论文笔记 MSCNN:A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection