项目目标是以有意义的方式分割图像。

The goal is to segment images in a meaningful way. 

伯克利收集了三百张图片，并向学生支付了每幅图片的手工分割费（通常每幅图片都会有多个手工分割）。

Berkeley collected three hundred images and paid students to hand-segment each one (usually each image has multiple hand-segmentations). 

其中200张是训练图像，其余100张是测试图像。

Two-hundred of these images are training images, and the remaining 100 are test images.

数据集包括用于读取图像和真实标注、计算基准测试分数以及一些其他实用功能的代码。

The dataset includes code for reading the images and ground-truth labels, computing the benchmark scores, and some other utility functions. 

它还包括用于图像分割示例的代码。

It also includes code for a segmentation example.

该数据集是比较新的，相关研究问题还没有解决，所以您有可能为您的项目设计出一个领先的算法。

This dataset is new and the problem unsolved, so there is a chance that you could come up with the leading algorithm for your project.

公开数据集下载地址：

https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/segbench/

项目思路：基于区域的图像分割Region-Based Segmentation

大多数图像分割算法都侧重于基于边缘或基于颜色和纹理的不连续性的分割。

Most segmentation algorithms have focused on segmentation based on edges or based on discontinuity of color and texture. 

然而，该数据集中的真实性允许有监督学习算法根据在某区域上计算的统计数据分割图像。

The ground-truth in this dataset, however, allows supervised learning algorithms to segment the images based on statistics calculated over regions. 

一种方法是将图像“过度分割”为超像素（Felzenszwalb 2004，代码可公开下载），并将超像素合并为更大的片段。

One way to do this is to "oversegment" the image into superpixels (Felzenszwalb 2004, code available) and merge the superpixels into larger segments. 

图形模型可以通过在相邻像素之间添加适当的势位来表示簇中的平滑度。

Graphical models can be used to represent smoothness in clusters, by adding appropriate potentials between neighboring pixels. 

在这个项目中，您可以展开的工作，例如，学习这样的势位，在具有非常大的树宽度模型中进行推理。

In this project, you can address, for example, learning of such potentials, and inference in models with very large tree-width.

更多精彩文章请关注微信号：