Object Identification, Segmentation, and Dimension Extraction from RGBD Images Using Deep Learning

This research proposes a comprehensive methodology for identifying, segmenting, and extracting dimensions of objects from RGBD images in real-time using deep learning techniques. The system integrates two primary components: a segmentation model for detecting and extracting objects, and a regression-based dimension extraction model to predict the dimensions of the detected objects. While the methodology is applicable to general objects, cylinders were chosen in this work.