Cascaded Hierarchical Model

Contextual information plays an important role in solving vision problems such as image segmentation. However, extracting contextual information and using it in an effective way remains a difficult problem. To address this challenge, we proposed a multi-resolution contextual framework, called cascaded hierarchical model (CHM), which learns contextual information in a hierarchical framework for image segmentation. At each level of the hierarchy, a classifier is trained based on downsampled input images and outputs of previous levels. Our model then incorporates the resulting multi-resolution contextual information into a classifier to segment the input image at original resolution. We repeat this procedure by cascading the hierarchical framework to improve the segmentation accuracy.

Multiple classifiers are learned in the CHM; therefore, a fast and accurate classifier is required to make the training tractable. The classifier also needs to be robust against overfitting due to the large number of parameters learned during training. So, the choice of the classifier is very crucial to the performance of this structure. Popular classifiers such as Random Forests and traditional artifical neural networks are not viable options. Neural networks are not good options mainly because of complexity, speed and average performance. Random Forests are not context-friendly structures and suffer from overfitting. LDNN is a perfect classification solution for CHM and we showed that CHM using LDNNs as classifiers is able to achieve state-of-the-art performance on many image segmentation problems.