Object detection is among the critical tasks in a wide range of real-world applications, especially in cases where there is a need to detect specific objects or situations. This study compares the performance of two state-of-the-art object detection models, YOLOv8 and YOLOv11, for classifying door states. Both models achieved an accuracy of 98.9% on the test dataset, though differences in precision and recall were observed. YOLOv8 demonstrated high accuracy in detection, which makes it ideal for tasks requiring precise classifications. YOLOv11 exhibited more stable performance across all door states, especially in ambiguous cases where the door’s state was near the threshold. This characteristic makes YOLOv11 more robust for real-world applications where confidence levels fluctuate. These findings underscore the potential for improving assistive technologies and robotic systems for visually impaired individuals by optimizing deep learning algorithms for accurate door state recognition.