Comprehensive Study and Analysis of 3D Objects from Real World Environment

Over this century, deep learning has undergone some remarkable furtherance in the understanding and analysis of 2D images. Regardless of these achievements, techniques for understanding and detecting 3D sensed data, from 2D images, is relatively infantile. This literature reviews existing state-of-the-art approaches of Deep Learning for detecting objects in 2D images, and detecting 3D objects from 2D images itself. We address the traditional methodologies as well as background concepts for object detection and image segmentation of 2D images eventually reviewing the latest vital approaches, comprising RGB-D, volumetric, and multi-view used for 3D objects. Datasets that are prominent for this process are also explained hereby. This paper represents a comparative analysis regarding the potential of deep learning for 3D sensed data along with its performance measures.