UAV-Assisted Logo Inspection: Deep Learning Techniques for Real- Time Detection and Classification of Distorted Logos

Ensuring the integrity of safety logos on aircraft is crucial for aviation personnel and overall safety. Presently, human operators perform inspections, which are susceptible to human errors. To address this, we propose an autonomous approach using drone-acquired photographic imagery for detecting and inspecting safety logos on fighter aircraft. Our methodology involves multiple stages: logo detection, distortion assessment, text orientation computation, and checking for logo overlap. We also calculate placement constraints for accurate logo positioning. We rigorously tested our approach on a local dataset, achieving an impressive precision of 92.3 % and a recall of 91.1 % for logo detection. We estimated computed text orientation in degrees and determined the distance between logos in pixels. This research presents a significant advancement in automatic logo inspection for aircraft safety. By leveraging drones and comprehensive detection techniques, our approach reduces human errors and enhances inspection efficiency. The potential impact includes improved safety standards in aviation and the foundation for future advancements in autonomous inspection systems.