Abstract

In this study, real-time detection and volume calculation of cargo packages on a conveyor belt were performed. The study was conducted using YOLOv8 and YOLOv11 models. The dataset, consisting of two classes, was labeled using the Labelme tool, and segmentation and region detection were performed. Packages were detected on the conveyor belt using the Intel RealSense D435i depth camera, and their volumes were accurately calculated. As a result of the training process, YOLOv11 was identified as the most successful model. YOLOv8 achieved mAP50: 0.993, mAP50-95: 0.956, precision: 0.978, and recall: 0.989, while YOLOv11 achieved mAP50: 0.995, mAP50-95: 0.976, precision: 1, and recall: 1. The results highlight the effectiveness of deep learning models, particularly the YOLO family, in object detection and volume measurement tasks in industrial environments.

References

[1] Zhou, Y., Liu, Z., & Xie, L. (2020). Deep learning applications in industrial automation: Object detection and segmentation. Journal of Industrial Automation, 34(3), 245-259.

[2] Redmon, J., & Farhadi, A. (2018). YOLOv4: Optimal speed and accuracy for real-time object detection. IEEE Conference on Computer Vision and Pattern Recognition, 1-9..

[3] Wang, S., Jiang, X., Zhao, J., Wang, X., Zhou, W., & Liu, Y. (2019, August). Vision based picking system for automatic express package dispatching. In 2019 IEEE international conference on real-time computing and robotics (RCAR) (pp. 797-802). IEEE.

[4] Wang, S., Jiang, X., Zhao, J., Wang, X., Zhou, W., & Liu, Y. (2019, August). Vision based picking system for automatic express package dispatching. In 2019 IEEE international conference on real-time computing and robotics (RCAR) (pp. 797-802). IEEE.

[5] Vina, A. (2025, Şubat 7). Package identification and segmentation with Ultralytics YOLO11. Ultralytics.Jan, R., Kour, H., Manhas, J., & Sharma, V. Recognition of Dry Fruits using Deep Convolutional Neural Network.

[6] Mohedano, M., Sánchez, D., & García, C. (2019). Real-time object detection and volume measurement with depth cameras in industrial automation. Journal of Industrial Robotics, 45(2), 123-135.