Object detection as a fundamental computer vision task has applications across different industries. One particular application is the detection and recognition of drawings, which often relies on convolutional neural networks. Developing a robust model for such an application typically requires a large dataset containing images of the target object from various perspectives and conditions. However, in many real-world cases, this is not possible, and the model must be trained on a small dataset. This paper addresses this challenging task by training a model for real-time detection of drawings using a constrained dataset. We leverage the state-of-the-art YOLO11 model as a foundation, known for its balance of speed and accuracy in real-time object detection. Different YOLO11 variants are trained under various configurations to evaluate their performance. The results of the study demonstrate the robustness and effectiveness of models for the real-time detection of drawings in cases when data resources are limited.
Science Journal 