Real-Time Pieces Detection and Recognition by the NAO Robot Using Tiny-YOLOv4/3L Algorithm

• Published in: 2024 International Symposium of Systems, Advanced Technologies and Knowledge (ISSATK) pp. 1 - 6
• Main Authors: Moetaz, Gharbi, Ibtissem, Chiha, Taoufik, Ladhari
• Format: Conference Proceeding
• Language: English
• Published: IEEE 02.11.2024
• Subjects: Accuracy; humanoid robot; Humanoid robots; Lighting; NAO Robot; Object detection; Real-time systems; Resilience; Robustness; Tiny-YOLOv4/3L algorithm
• DOI: 10.1109/ISSATK62463.2024.10808621

This research paper presents a system for object detection using the Tiny-YOLOv4/3L model, with a focus on a humanoid robot NAO. The system is designed to detect specific objects under various conditions. The efficacy of the model was evaluated based on its performance under standard conditions and under common real-world scenarios such as occlusion and low light. The results demonstrate that the model can provide reliable detection with commendable accuracy under standard conditions and shows a degree of resilience against minor occlusion and low light. However, significant occlusion and dim light conditions pose considerable challenges to the performance of the model.