Adaptive Indoor Scene Classification with Multi-SVM Classification to solve Multi-Class Problem

In recent years, indoor scene recognition has attracted much attention and its research has rapidly expanded by not only engineers but also neuroscientists, since it has many potential applications in computer vision communication and automatic access control system. Especially, indoor scene recognition is an important part of computer vision and feature recognition as the first step of automatic uninterrupted robotic movement or computer vision applications like automatic interior designing algorithms. However, the indoor scene detection is not straightforward because it has lots of variations of image appearance, such as light effect, occlusion, image orientation, illuminating condition and object variety. Many novel methods have been proposed to resolve each variation listed above. For example, the template-matching methods are used for indoor scene localization and detection by computing the correlation of an input image to a standard and training scene appearance or pattern. The feature invariant approaches are used for feature detection of bed, chair, cabinet, table, door, electrical or electronic items, etc. The appearance-based methods are used for indoor feature detection with support vector machine and information theoretical approach. Nevertheless, implementing the methods altogether is still a great challenge. Fortunately, the images used in this project have some degree of uniformity thus the detection algorithm can be simpler: first, the all the faces are vertical and have frontal view; second, they are under almost the same illuminate condition. This project presents an indoor scene detection technique mainly based on the appearance based feature segmentation, SVM training and SVM classification methods to recognize the indoor scenes.