Damaged Parcel Detection in the Logistics Sector Using the Yolov12 Model with Roboflow 3.0

Abstract

This study presents an automated system for detecting physical damage to packages in logistics operations using deep learning. Manual inspections are often slow and error prone, whereas deep learning based object detection, particularly the YOLO (You Only Look Once) family, offers both speed and accuracy for real time applications. We created a dataset of 5,188 images labelled as “Damaged” or “Undamaged” on the Roboflow 3.0 platform. To enhance the robustness, each image was augmented with greyscale conversion and 90° rotations (clockwise and counterclockwise). The dataset was divided into 86% training, 7% validation, and 7% test sets and used to train the YOLOv12 model. The model achieved a mean average precision at IoU = 0.50 (mAP@0.50) of 93.7%, with a Precision of 93.3% and a Recall of 93.8% on the test set. Integrated into a Python GUI, the model processes user selected images in real time, drawing bounding boxes around packages and labelling their damage status. This approach significantly improves the operational efficiency and detection accuracy by reducing the reliance on manual inspections. Future work will investigate advanced augmentation techniques (e.g., varying lighting, contrast, and shadows) and transfer learning to enhance generalisation across diverse package types and environmental conditions

Keywords

• Object Detection
• Artificial Intellegence
• Deep Learning
• Logistics

References

1. Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. 2016. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. https://doi.org/10.48550/arXiv.1506.01497 google scholar
2. Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. 2017. ImageNet classification with deep convolutional neural networks. Commun. ACM 60, 6 (May 2017), 84–90. https://doi.org/10.1145/3065386 google scholar
3. Connor Shorten and Taghi M. Khoshgoftaar. 2019. A survey on Image Data Augmentation for Deep Learning. J Big Data 6, 1 (December 2019), 60. https://doi.org/10.1186/s40537-019-0197-0 google scholar
4. Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, and Andrew Rabinovich. 2015. Going deeper with convolutions. In 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2015. IEEE, Boston, MA, USA, 1–9. https://doi.org/10.1109/CVPR.2015.7298594 google scholar
5. Li Deng and Dong Yu. 2014. Deep Learning: Methods and Applications. FNT in Signal Processing 7, 3–4 (2014), 197–387. https://doi. org/10.1561/2000000039 google scholar
6. M. F. Keskenler and E. F. Keskenler. 2017. Geçmişten Günümüze Yapay Sinir Ağları ve Tarihçesi. Takvim-I Vekayi 5, 2 (2017), 8–18. google scholar
7. Yunjie Tian, Qixiang Ye, and David Doermann. 2025. YOLOv12: Attention-Centric Real-Time Object Detectors. https://doi.org/10. 48550/arXiv.2502.12524 google scholar
8. Mujadded Al Rabbani Alif. 2024. YOLOv11 for Vehicle Detection: Advancements, Performance, and Applications in Intelligent Transportation Systems. https://doi.org/10.48550/arXiv.2410.22898 google scholar

9. Mujadded Al Rabbani Alif and Muhammad Hussain. 2025. YOLOv12: A Breakdown of the Key Architectural Features. https://doi. org/10.48550/ARXIV.2502.14740 google scholar
10. Mehmet Nergıi̇z. 2023. Enhancing Strawberry Harvesting Efficiency through Yolo-v7 Object Detection Assessment. Turkish Journal of Science and Technology 18, 2 (September 2023), 519–533. https://doi.org/10.55525/tjst.1342555 google scholar
11. Kadir Diler Alemdar. 2024. Çift Sıra Parklanma Durumunun Nesne Tespit Algoritması YOLOv8 ile Tespit Edilmesi. Iğdır Üniversitesi Fen Bilimleri Enstitüsü Dergisi 14, 3 (September 2024), 1164–1176. https://doi.org/10.21597/jist.1472194 google scholar
12. Batuhan Karadağ and Ali Ari. 2023. Akıllı Mobil Cihazlarda YOLOv7 Modeli ile Nesne Tespiti. Politeknik Dergisi 26, 3 (October 2023), 1207–1214. https://doi.org/10.2339/politeknik.1296541 google scholar
13. Stanley Glenn E. Brucal, Luigi Carlo M. De Jesus, and Leonardo A. Samaniego. 2024. Development of a Localized Tomato Leaf Disease Detection using YoloV9 Model via RoboFlow 3.0. In 2024 IEEE 13th Global Conference on Consumer Electronics (GCCE), October 29, 2024. IEEE, Kitakyushu, Japan, 601–603. https://doi.org/10.1109/GCCE62371.2024.10760343 google scholar
14. Franklin G. Lopez and Arnel C. Fajardo. 2024. Cacao Health Classification using RoboFlow Object Detection Model. In 2024 7th Inter^ national Seminar on Research of Information Technology and Intelligent Systems (ISRITI), December 11, 2024. IEEE, Yogyakarta, Indonesia, 242–246. https://doi.org/10.1109/ISRITI64779.2024.10963535 google scholar
15. David M. W. Powers. 2020. Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. https://doi.org/10.48550/arXiv.2010.16061 google scholar
16. Mark Everingham, Luc Van Gool, Christopher K. I. Williams, John Winn, and Andrew Zisserman. 2010. The Pascal Visual Object Classes (VOC) Challenge. Int J Comput Vis 88, 2 (June 2010), 303–338. https://doi.org/10.1007/s11263-009-0275-4 google scholar