Retina Net–YOLOv8 Hybrid Multispectral UAV Framework with Unsupervised Segmentation for Agricultural Weed and Pest Detection
Keywords:
Precision Agriculture, RetinaNet, YOLOv8, Multispectral UAV, Weed Detection, Pest Identification, Unsupervised Segmentation, IoT, Deep LearningAbstract
We present a RetinaNet–YOLOv8 hybrid multispectral UAV framework integrated with unsupervised segmentation for real-time weed and pest detection in precision agriculture. The system leverages high-resolution RGB and multispectral imagery captured by UAV platforms, enabling enhanced vegetation–background discrimination through spectral index computation (NDVI, GNDVI). The RetinaNet module, optimized for high-accuracy detection using focal loss, achieved a mean Average Precision (mAP) of 0.947, while YOLOv8 delivered ultra-fast inference at 38 FPS with minimal accuracy trade-off (mAP = 0.944). An unsupervised segmentation component based on RoWeeder attained an F1-score of 75.3%, reducing annotation requirements and accelerating deployment in data-scarce environments. Additionally, an AI–IoT pest monitoring subsystem provided early infestation alerts up to three months ahead of conventional scouting methods. Benchmarking against U-Net and DETR demonstrated that the proposed hybrid approach offers superior detection accuracy, faster inference, and robust field performance. This integrated solution represents a scalable, cost-effective, and edge-deployable framework for sustainable agricultural weed and pest management.
Downloads
References
[1]. Siddiqui, S.A., et al. (2021). "Neural Network based Smart Weed Detection System." IEEE Xplore.
[2]. Junior, L.C.M., et al. (2021). "Real Time Weed Detection using Computer Vision and YOLOv5." IEEE Xplore.
[3]. Vivek, K.K., et al. (2021). "Pests & weed control autonomous robot using machine learning." IEEE Xplore.
[4]. Li, S., et al. (2025). "PD-YOLO: A novel weed detection method based on multi-scale feature fusion." Frontiers Plant Science.
[5]. Pai, D.G., et al. (2024). "Deep Learning Techniques for Weed Detection in Agriculture." IEEE Xplore.
[6]. Chithambarathanu, M., et al. (2023). "Survey on crop pest detection using deep learning and computer vision." PMC.
[7]. Goel, A., et al. (2020). "Multispectral UAV Imagery for Weed Identification." IEEE Xplore.
[8]. Siddiqui, S.A., et al. (2021). "CNN for Early Detection of Weeds." AWS Journal.
[9]. Rai, N., et al. (2023). "Applications of deep learning in precision weed management." ScienceDirect.
[10]. Meena, H., et al. (2023). "Deep Learning for Invasive Weed Species Classification." AWS Journal.
[11]. Razfar, M., et al. (2022). "Weed Detection System using CNN and DL Models." AWS Journal.
[12]. Nasiri, S., et al. (2022). "Pixel-wise Segmentation of Weeds, Soil, and Sugar Beet." AWS Journal.
[13]. Shorewala, R., et al. (2021). "Weed Density Estimation using Deep Semi-supervised Learning." AWS Journal.
[14]. Alrowais, A., et al. (2022). "IoT based Weed Recognition and Classification." AWS Journal.
[15]. Wang, P., et al. (2022). "Weed25: A Deep Learning Dataset for Weed Identification." AWS Journal.
[16]. Adhinata, F.D., et al. (2024). "A comprehensive survey on weed and crop classification using machine learning and deep learning." ScienceDirect.
[17]. Gomes, G.F., et al. (2022). "Machine learning algorithms applied to weed management in integrated crop-livestock systems." AWS Journal.
[18]. Chavan, M.S. & Nandedkar, A.V. (2018/2021). "CNN for Automatic Weed Classification in Farming." AWS Journal.
[19]. Chen, J., et al. (2018/2021). "High-resolution Multispectral Satellite Data for Wheat Rust Disease." PMC.
[20]. Deng, G., et al. (2018/2021). "Image Insect Pest Surveillance with SVM." PMC.
[21]. Precision Agriculture Crop Recommendation System Using IoT and ML. (2023). IEEE Xplore.
[22]. Multispectral Remote Sensing for Weed Detection in West Australian Farms. (2025). IEEE Xplore.
[23]. Lightweight Deep Learning Model for Weed Detection for IoT Devices. (2022). IEEE Xplore.
[24]. Weed Identification and Removal: Deep Learning Techniques and Research Advancements. (2022). IEEE Xplore.
[25]. Pest Detection using Machine Learning. (2024). IEEE Xplore.
