Machine Learning-Based Indoor Scenes Identification Using Thepade’s SBTC and Local Binary Patterns

Abstract

AbstractNow day’s images are an essential thing in human life. The images are get captured everywhere in the world. Scene classification is one of the important problems. The images are taken now and then at many different places. These different places have many different scenes like airport, bakery, kitchen, mountain, etc. Scene classification is mostly in two forms, which are indoor scene classification and outdoor scene classification. A difficulty in scene classification causes problems in the research and image-processing field. The image classification gets difficult because of the many distinct objects where present in the different sceneries, this makes a machine get confused about the categories of the scene. The present problem for the scene classification gives poor accuracy for classification. The object varies in different shapes, sizes in numerous locations, which is also one of the reasons behind the poor accuracy of the indoor scene classification. Here, we are going to perform feature extraction for pattern recognition and different machine learning models for image classification using the MIT Indoor Scene dataset. Here, paper gives Thepade's SBTC n-ary feature extraction method on RGB and LUV color plane models alongside LBP. The comparison of the given three feature extraction methods Thepade's SBTC with RGB color plane has given better accuracy.KeywordsIndoor scene classificationFeature extractionThepade's SBTCLocal Binary Pattern