Web-based Timber Logs Information System Using the YOLOv8 Model: IstifTakip
Abstract
This study introduces İstifTakip, a web-based information system developed for the automated detection and measurement of stacked timber logs using the YOLOv8 deep learning model. The system aims to overcome the limitations of manual timber measurement methods by providing a more accurate and efficient alternative. Data were collected through smartphone images of timber stacks at the Ulucak Forest Depot in İzmir, Türkiye. The YOLOv8 model, optimized using the Optuna library, was trained on this dataset to detect logs and calculate key attributes such as diameter and volume. Hyperparameter optimization with Optuna resulted in a significant improvement in model performance, achieving an mAP@0.5 score of 0.8569, precision of 0.8513, and recall of 0.8827. These results demonstrate the model’s robustness and accuracy in detecting logs across varied image conditions. İstifTakip was developed using the Django framework and offers a user-friendly interface where users can upload images, annotate reference lines, and obtain log measurements. The system is specifically designed for Turkish forestry, supporting local language and practices, which sets it apart from other global solutions. Its scalability and potential for integration with mobile devices make it a valuable tool for future forestry applications. This research highlights the advantages of combining deep learning and smart forestry technologies to enhance operational efficiency and data accuracy in timber stack management.
Keywords
Kaynakça
- Acar, H. (2004). Lecture notes on transportation in forestry (Original in Turkish). Karadeniz Technical University, Faculty of Forestry, Department of Forest Engineering.
- Berendt, F., Wolfgramm, F., & Cremer, T. (2021). Reliability of photo-optical measurements of log stack gross volume. Silva Fennica, 55(3), Article 10555.
- Buğday, S. E., & Özden, S. (2015). TSE standards in the forest products market in Turkey (Original in Turkish). In D. Toksoy & M. M. Bayramoğlu (Eds.), IV Symposium on Socio-Economic Issues in Forestry (pp. 285–297). Karadeniz Technical University, Faculty of Forestry, Department of Forest Economics.
- Casas, G. G., Ismail, Z. H., Limeira, M. M. C., da Silva, A. A. L., & Leite, H. G. (2023). Automatic detection and counting of stacked eucalypt timber using the YOLOv8 model. Forests, 14(12), Article 2369.
- Dietz, H. U., Jensen, L., & Seeling, U. (2019). Measuring log piles with photo-optical mono camera systems. In Proceedings of the 52nd International Symposium on Forestry Mechanization (FORMEC) - Exceeding the Vision: Forest Mechanisation of the Future (p. 302). Sopron, Hungary.
- Eker, R., & Aydın, A. (2020). The use of unmanned aerial vehicle (UAV) for tracking stock movements in forest enterprise depots. European Journal of Forest Engineering, 6(2), 68–77.
- Gül, A. U., & Özçelik, R. (2017). Minimizing production costs in forest enterprises using 0-1 integer linear programming (Original in Turkish). In S. Özden (Ed.), IV National Forestry Congress "Human-Nature Interaction in Forest and Forestry", Vol. 2 (pp. 183–192). Turkish Foresters' Association Publications.
- Karha, K., Nurmela, S., Karvonen, H., Kivinen, V. P., Melkas, T., & Nieminen, M. (2019). Estimating the accuracy and time consumption of a mobile machine vision application in measuring timber stacks. Computers and Electronics in Agriculture, 158, 167–182.
- Köse, N., Akkemik, Ü., & Güner, T. (2017). Some old black pine stands in Anatolia that need to be protected (Original in Turkish). In S. Özden (Ed.), IV National Forestry Congress "Human-Nature Interaction in Forest and Forestry", Vol. 2 (pp. 339–349). Turkish Foresters' Association Publications.
- Purfürst, T., de Miguel-Díez, F., Berendt, F., Engler, B., & Cremer, T. (2022). Comparison of wood stack volume determination between different manual, photo-optical, iPad-LiDAR, and handheld-LiDAR-based measurement methods.