This report proposes a new model using the Multi-class Multi-scale Stacked Sequential Learning framework as the solution to the problem of indoor semantic segmentation. Following recent trends in state-of-the-art, a base classifier uses an initial SLIC segmentation to obtain superpixels which provide a diminution of data while retaining object boundaries. A series of color and depth features are extracted from the superpixels and are used in a Conditional Random Field to predict superpixel labels. Furthermore, a Random Forest classifier using random offset features is also used as an input to the Conditional Random Field, acting as an initial prediction.
As a stacked classifier, another Random Forest is used which acts on a spatial multi-scale decomposition of the confidence map to correct the erroneous labels assigned by the previous classifier.
The model is tested and trained on the popular NYU-v2 dataset. The approach shows that simple features with the power of the MMSSL framework can achieve better performance than similar methods.