• Version
• Download 27
• File Size 616.98 KB
• File Count 1
• Create Date 7 December 2025
• Last Updated 7 December 2025

Custom Object Detection Using YOLOV11 Model

 

Ruchika Mittal1, Devender Singh Topwal2
Department of Electrical & Electronics Engineering

Bhagwan Parshuram Institute of Technology GGSIPU, New Delhi-110036

 

ABSTRACT
The evolution of deep learning has revolutionized visual recognition, yet traditional detection models often struggle with the trade-off between accuracy and inference speed in custom scenarios. This paper presents a novel implementation of the YOLOv11 model tailored for custom object detection tasks. Utilizing a transfer learning methodology, the framework effectively adapts high-level feature representations from pre-trained weights to a custom dataset, ensuring adaptability across diverse visual variations. The investigation highlights the architectural advancements of YOLOv11, including its optimized backbone and unified detection pipeline. Quantitative analysis using mean Average Precision (mAP), Precision, and Recall metrics demonstrates the model's efficacy in minimizing false negatives while maintaining real-time performance. This study confirms YOLOv11’s viability as a lightweight, high-performance architecture suitable for deployment in resource-constrained computer vision systems.

Keywords: YOLOv11, Object Detection, Deep Learning, Computer Vision, Real-Time Detection, Custom Dataset, Segmentation, mAP.

Download