Comparative Analysis of Deep Learning Architectures for Coffee Tree Detection from Aerial Imagery
DOI:
https://doi.org/10.30871/jaic.v10i2.12360Keywords:
Deep Learning, YOLO, Aerial Images, AugmentationAbstract
Coffee cultivation plays a vital economic role globally, supporting millions of livelihoods. Traditional manual enumeration methods for crop monitoring are time-intensive, costly, and prone to errors, particularly on large-scale farms. This study addresses the need for automated coffee tree detection systems by systematically evaluating five state-of-the-art deep learning architectures: YOLOv8 (nano, small, medium), Faster R-CNN, and EfficientDet. Using a dataset of 1,500 high-resolution aerial images from coffee plantations in Lampung, we investigated four critical aspects: optimal object detection architecture, effective augmentation strategies, minimum data requirements, and error patterns. Results demonstrate that YOLOv8n achieves superior performance with 95.98% [email protected], outperforming larger variants and two-stage detectors. Basic augmentation techniques proved most effective, with [email protected] of 96.13%, surpassing aggressive strategies like mosaic and mixup that disrupted the spatial structure of the plantations. Data efficiency analysis revealed that 750 images (50% of the dataset) achieved 99.55% of peak performance, enabling cost-effective deployment in resource-constrained scenarios. Error analysis indicated that false positives were the primary challenge, which is addressable through confidence threshold calibration. These findings provide evidence-based guidelines for practitioners, demonstrating that compact architectures with moderate augmentation can achieve high accuracy with limited data, facilitating the practical deployment of precision agriculture technologies in coffee cultivation.
Downloads
References
[1] S. Krishnan, “Sustainable Coffee Production,” Oxford Research Encyclopedia of Environmental Science, Jun. 2017, doi: https://doi.org/10.1093/acrefore/9780199389414.013.224
[2] X. Wang, C. Zhang, Z. Qiang, C. Liu, X. Wei, and F. Cheng, “A Coffee Plant Counting Method Based on Dual-Channel NMS and YOLOv9 Leveraging UAV Multispectral Imaging,” Remote Sensing, vol. 16, no. 20, pp. 3810–3810, Oct. 2024, doi: https://doi.org/10.3390/rs16203810
[3] J. Bolaños, Juan Carlos Corrales, and Liseth Viviana Campo, “Feasibility of Early Yield Prediction per Coffee Tree Based on Multispectral Aerial Imagery: Case of Arabica Coffee Crops in Cauca-Colombia,” Remote Sensing, vol. 15, no. 1, pp. 282–282, Jan. 2023, doi: https://doi.org/10.3390/rs15010282.
[4] F. Rovira-Más and Verónica Saiz-Rubio, “Crop Scouting and Surrounding Awareness for Specialty Crops,” Agriculture automation and control, pp. 111–136, Jan. 2021, doi: https://doi.org/10.1007/978-3-030-70400-1_5.
[5] C. Bunn, P. Läderach, O. Ovalle Rivera, and D. Kirschke, “A bitter cup: climate change profile of global production of Arabica and Robusta coffee,” Climatic Change, vol. 129, no. 1–2, pp. 89–101, Dec. 2014, doi: https://doi.org/10.1007/s10584-014-1306-x.
[6] A. F. Colaço and J. P. Molin, “Variable rate fertilization in citrus: a long term study,” Precision Agriculture, vol. 18, no. 2, pp. 169–191, May 2016, doi: https://doi.org/10.1007/s11119-016-9454-9.
[7] W. H. Maes and K. Steppe, “Perspectives for Remote Sensing with Unmanned Aerial Vehicles in Precision Agriculture,” Trends in Plant Science, vol. 24, no. 2, pp. 152–164, Feb. 2019, doi: https://doi.org/10.1016/j.tplants.2018.11.007.
[8] Y. Ampatzidis and V. Partel, “UAV-Based High Throughput Phenotyping in Citrus Utilizing Multispectral Imaging and Artificial Intelligence,” Remote Sensing, vol. 11, no. 4, p. 410, Feb. 2019, doi: https://doi.org/10.3390/rs11040410.
[9] A. Kamilaris and F. X. Prenafeta-Boldú, “Deep learning in agriculture: A survey,” Computers and Electronics in Agriculture, vol. 147, pp. 70–90, Apr. 2018, doi: https://doi.org/10.1016/j.compag.2018.02.016.
[10] L. Tang and G. Shao, “Drone remote sensing for forestry research and practices,” Journal of Forestry Research, vol. 26, no. 4, pp. 791–797, Jun. 2015, doi: https://doi.org/10.1007/s11676-015-0088-y.
[11] L. S. Santana, G. A. e S. Ferraz, G. H. R. dos Santos, N. L. Bento, and R. de O. Faria, “Identification and Counting of Coffee Trees Based on Convolutional Neural Network Applied to RGB Images Obtained by RPA,” Sustainability, vol. 15, no. 1, p. 820, Jan. 2023, doi: https://doi.org/10.3390/su15010820.
[12] R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation,” openaccess.thecvf.com, 2014. https://openaccess.thecvf.com/content_cvpr_2014/html/Girshick_Rich_Feature_Hierarchies_2014_CVPR_paper.html
[13] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” Cv-foundation.org, pp. 779–788, 2016, Available: https://www.cv-foundation.org/openaccess/content_cvpr_2016/html/Redmon_You_Only_Look_CVPR_2016_paper.html
[14] M. Tan, R. Pang, and Q. V. Le, “EfficientDet: Scalable and Efficient Object Detection,” openaccess.thecvf.com, 2020. https://openaccess.thecvf.com/content_CVPR_2020/html/Tan_EfficientDet_Scalable_and_Efficient_Object_Detection_CVPR_2020_paper.html
[15] C. Shorten and T. M. Khoshgoftaar, “A survey on Image Data Augmentation for Deep Learning,” Journal of Big Data, vol. 6, no. 1, Jul. 2019, doi: https://doi.org/10.1186/s40537-019-0197-0.
[16] Z. Khan, Y. Shen, and H. Liu, “ObjectDetection in Agriculture: A Comprehensive Review of Methods, Applications, Challenges, and Future Directions,” Agriculture, vol. 15, no. 13, p. 1351, Jun. 2025, doi: https://doi.org/10.3390/agriculture15131351.
[17] H. Raza, M. A. Bakr, S. D. Khan, Hira Batool, H. Ullah, and M. Ullah, “Benchmarking YOLO Models for Crop Growth and Weed Detection in Cotton Fields,” AgriEngineering, vol. 7, no. 11, pp. 375–375, Nov. 2025, doi: https://doi.org/10.3390/agriengineering7110375.
[18] A. Kamilaris and Prenafeta-Boldú, Francesc X, “Disaster Monitoring using Unmanned Aerial Vehicles and Deep Learning,” arXiv.org, 2024. https://arxiv.org/abs/1807.11805 (accessed Oct. 01, 2024).
[19] J. G. M. Esgario, R. A. Krohling, and J. A. Ventura, “Deep learning for classification and severity estimation of coffee leaf biotic stress,” Computers and Electronics in Agriculture, vol. 169, p. 105162, Feb. 2020, doi: https://doi.org/10.1016/j.compag.2019.105162.
[20] E. Silva, J. B. Fragoso, Thuanne Paixão, A. B. Alvarez, and Facundo Palomino-Quispe, “A Low Computational Cost Deep Learning Approach for Localization and Classification of Diseases and Pests in Coffee Leaves,” IEEE Access, pp. 1–1, Jan. 2025, doi: https://doi.org/10.1109/access.2025.3562832.
[21] M. A. Tamayo-Monsalve et al., “Coffee Maturity Classification Using Convolutional Neural Networks and Transfer Learning,” IEEE Access, vol. 10, pp. 42971–42982, 2022, doi: https://doi.org/10.1109/access.2022.3166515.
[22] D. Su, H. Kong, Y. Qiao, and S. Sukkarieh, “Data augmentation for deep learning based semantic segmentation and crop-weed classification in agricultural robotics,” Computers and Electronics in Agriculture, vol. 190, p. 106418, Nov. 2021, doi: https://doi.org/10.1016/j.compag.2021.106418.
[23] Z. Zou, Z. Shi, Y. Guo, and J. Ye, “Object Detection in 20 Years: A Survey,” arXiv.org, 2019. https://arxiv.org/abs/1905.05055
[24] L. Liu et al., “Deep Learning for Generic Object Detection: A Survey,” arXiv:1809.02165 [cs], Aug. 2019, Available: https://arxiv.org/abs/1809.02165
[25] N. Dalal and B. Triggs, “Histograms of Oriented Gradients for Human Detection,” 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), vol. 1, pp. 886–893, 2005, doi: https://doi.org/10.1109/cvpr.2005.177.
[26] P. F. Felzenszwalb, R. B. Girshick, D. McAllester, and D. Ramanan, “Object Detection with Discriminatively Trained Part-Based Models,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 32, no. 9, pp. 1627–1645, Sep. 2010, doi: https://doi.org/10.1109/tpami.2009.167.
[27] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Advances in Neural Information Processing Systems, vol. 25, pp. 1097–1105, 2012, Available: https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html
[28] W. Liu et al., “SSD: Single Shot MultiBox Detector,” Computer Vision – ECCV 2016, vol. 9905, no. 5, pp. 21–37, 2016, doi: https://doi.org/10.1007/978-3-319-46448-0_2.
[29] R. Girshick, “Fast R-CNN,” openaccess.thecvf.com, 2015. https://openaccess.thecvf.com/content_iccv_2015/html/Girshick_Fast_R-CNN_ICCV_2015_paper.html
[30] J. Redmon and A. Farhadi, “YOLO9000: Better, Faster, Stronger,” openaccess.thecvf.com, 2017. https://openaccess.thecvf.com/content_cvpr_2017/html/Redmon_YOLO9000_Better_Faster_CVPR_2017_paper.html
[31] A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Optimal Speed and Accuracy of Object Detection,” arXiv, vol. 1, Apr. 2020, Available: https://arxiv.org/abs/2004.10934
[32] R. Varghese and Sambath M, “YOLOv8: A Novel Object Detection Algorithm with Enhanced Performance and Robustness,” IEEE, Apr. 2024, doi: https://doi.org/10.1109/adics58448.2024.10533619.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Alya Khairunnisa Rizkita, Andre Febrianto, Miranti Verdiana, Amirul Iqbal, Muhammad Habib Algifari
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License (Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
