Real Time Bolt Segmentation on Diverse Backgrounds using Deep Learning
G. Ramesh Chandra, Madhavi Batthala, Sathwika Ega, Harshitha Mullagiri, Srija Nomula · 2024
Bolts are vital components used in various structures to ensure stability and safety. Detecting bolts accurately is essential for structural health monitoring and maintenance across diverse backgrounds and environments. This paper aims to develop a real-time mobile application capable of robustly segmenting bolts in different settings using deep learning techniques. The proposed solution leverages the YOLO (You Only Look Once) framework, a powerful deep learning algorithm known for real-time object detection. This application holds potential for applications in industries such as construction, infrastructure inspection, and maintenance, ensuring the integrity and safety of structures in a wide range of scenarios. The proposed mobile based solution is able to detect the bolts with an mean average precision of 92.7%.