DEVELOPMENT OF A YOLOv۸ MODEL FOR DOORDETECTION USING ARTIFICIAL INTELLIGENCE ANDPYTORCH

Doors serve as crucial gateways in our built environment, and theirprecise detection is essential for various applications, including robotics,autonomous navigation, and security systems. However, accuratelydetecting doors in diverse real-world settings poses challenges. Thispaper delves into the realm of door detection using deep learningtechniques, specifically focusing on the YOLOv۸ algorithm. The papercommences by highlighting the significance of door detection and thelimitations of traditional methods. It then explores YOLOv۸, a singlestageobject detection algorithm known for its speed and efficiency. Italso describes the chosen model configuration, training parameters, anddataset employed. The dataset consists of over ۴۷۰ images withcorresponding labels, and the model is trained using a pre-trainedYOLOv۸s model for faster convergence. Visualizations comparingpredicted door bounding boxes with original images are also included.Overall, this paper demonstrates the potential of YOLOv۸ for doordetection in diverse real-world scenarios. While further improvementsare possible, the presented approach offers a promising solution with itsspeed, efficiency, and adaptability.

Keywords:

Door Detection , Deep Neural Network , YOLOv۸ , PYTORCH