Vision Based Object Dimension Detection
Shripad S. Bhatlawande, Swati Shilaskar, Prathamesh Kachkure, Prathamesh Dhole, Devyani Patil · 2023
This paper presents a method for real-time object dimension detection of commonly used objects: pens, wallets, and keys. This system is ported on Jetson Nano. The results are translated as an audio output which helps especially the visually impaired individuals to identify the objects. This methodology involves data collection of the objects, preprocessing of the dataset. YOLO (You Only Look Once) model is trained and tested using the collected dataset. The deployment of low power embedded system is a novel approach of object detection and dimension detection in real-time with the use of Convolutional neural network algorithm. The system is efficient in detecting dimension with an accuracy of 84.35% for pen, 82.07% for key and 87.84% with dimensional variance of ±0.29 cm in length and ± 0.18 cm in width.