AgroYOLOvNext: A Multi-Scale Transformer-Enhanced YOLO Framework with Domain-Adaptive Learning for Real-Time Plant Disease Detection under Field Conditions
Main Article Content
Abstract
Automated plant disease detection in a real-world agricultural setting remains a difficult problem because of variation in illumination, intra-class diversity, domain shift between laboratory and field data, and occlusions. While recent deep learning-based object detection models, such as YOLO variants and transformer-based models, have demonstrated impressive accuracy on curated datasets, their ability to perform well under open field conditions remains limited. To address this, we introduce Agro YOLO v Next, a multi-scale Transformer-enhanced YOLO framework with domain-adaptive learning that combines three complementary components rather than a single new mechanism. The proposed architecture consists of three main components: (1) a Multi-Scale Depth wise Asymmetric (MSDA) backbone for extracting fine-grained features from multiple receptive fields, (2) a Lightweight Cross-Stage Transformer Neck (LCST-Neck) for improved feature fusion and class separability, and (3) a Domain-Adaptive Contrastive Head (DACH) to encourage feature representations that are more invariant to common imaging perturbations. All experiments were conducted on the laboratory-acquired Plant Village dataset (54,306 images, 38 classes); no field-acquired imagery was used for training or evaluation. All models, including YOLO-family detectors adapted with a global-average-pooling classification head, are evaluated under a common whole-image classification protocol (accuracy, precision, recall, F1-score, AUC); no bounding-box annotations, IoU, or mAP are used, and this comparison basis is stated explicitly so that CNN classifiers and detector backbones are judged on equal footing. Under this protocol, the proposed model achieves an accuracy of 98.34%, an AUC of 0.998, a precision of 98.11%, a recall of 97.89%, and an F1-score of 98.00%, outperforming the strongest of seven baseline architectures (YOLOv9-E) by 1.40 accuracy percentage points. These results indicate the model is effective and competitive on the Plant Village benchmark; claims of real-world field generalization are not yet supported by field-acquired data and should be regarded as a direction for future validation rather than a demonstrated outcome.
Article Details
Section
This work is licensed under a Creative Commons Attribution 4.0 International License.
How to Cite
References
[1] FAO, "New standards to curb the global spread of plant pests and diseases," Food and Agriculture Organization of the United Nations, 2019. [Online]. Available: https://www.fao.org/plant-production-protection/about/en
[2] S. P. Mohanty, D. P. Hughes, and M. Salathé, "Using deep learning for image-based plant disease detection," Frontiers in Plant Science, vol. 7, p. 1419, Sep. 2016, doi: 10.3389/fpls.2016.01419.
[3] Y. Miao, W. Meng, and X. Zhou, "SerpensGate-YOLOv8: an enhanced YOLOv8 model for accurate plant disease detection," Frontiers in Plant Science, vol. 15, p. 1514832, Jan. 2025, doi: 10.3389/fpls.2024.1514832.
[4] X. Wu, X. Fan, P. Luo, S. D. Choudhury, T. Tjahjadi, and C. Hu, "From laboratory to field: unsupervised domain adaptation for plant disease recognition in the wild," Plant Phenomics, vol. 5, p. 0038, Mar. 2023, doi: 10.34133/plantphenomics.0038.
[5] P. Kalpana and R. Anandan, "Plant disease recognition using residual convolutional enlightened Swin transformer networks," Scientific Reports, vol. 14, p. 8660, Apr. 2024, doi: 10.1038/s41598-024-56393-8.
[6] J. Mu, Q. Feng, J. Yang, J. Zhang, and S. Yang, "Few-shot disease recognition algorithm based on supervised contrastive learning," Frontiers in Plant Science, vol. 15, p. 1341831, Feb. 2024, doi: 10.3389/fpls.2024.1341831.
[7] Z. Liu et al., "Swin Transformer: Hierarchical vision transformer using shifted windows," in Proc. IEEE/CVF Int. Conf. Comput. Vis. (ICCV), Montreal, Canada, Oct. 2021, pp. 9992–10002, doi: 10.1109/ICCV48922.2021.00986.
[8] M. Xu, S. Yoon, Y. Jeong, and D. S. Park, "Transfer learning for versatile plant disease recognition with limited data," Frontiers in Plant Science, vol. 13, p. 1010981, Nov. 2022, doi: 10.3389/fpls.2022.1010981.
[9] F. Barbosa and R. Santos, "Transformer-based architectures for agricultural remote sensing: a systematic review, 2022–2024," Remote Sensing, vol. 16, no. 3, p. 512, Jan. 2024, doi: 10.3390/rs16030512.
[10] A. Muhammad, Z. Salman, K. Lee, and D. Han, "Harnessing the power of diffusion models for plant disease image augmentation," Frontiers in Plant Science, vol. 14, p. 1280496, Nov. 2023, doi: 10.3389/fpls.2023.1280496.
[11] G. Hao, "Knowledge distillation facilitates the lightweight and efficient plant diseases detection model," Plant Phenomics, vol. 5, p. 0062, 2023, doi: 10.34133/plantphenomics.0062.
[12] M. A. Noyan, "Uncovering bias in the PlantVillage dataset," arXiv preprint arXiv:2206.04374, Jun. 2022. [Online]. Available: https://arxiv.org/abs/2206.04374
[13] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, "Grad-CAM: visual explanations from deep networks via gradient-based localization," Int. J. Comput. Vis., vol. 128, pp. 336–359, Oct. 2020, doi: 10.1007/s11263-019-01228-7.
[14] P. Khosla et al., "Supervised contrastive learning," in Advances in Neural Information Processing Systems (NeurIPS), vol. 33, 2020, pp. 18661–18673. [Online]. Available: https://arxiv.org/abs/2004.11362
[15] X. Lu, T. Gong, and X. Li, "CLA: a self-supervised contrastive learning method for leaf disease identification with domain adaptation," Comput. Electron. Agric., vol. 211, p. 107967, 2023, doi: 10.1016/j.compag.2022.107967
[16] I. Loshchilov and F. Hutter, "Decoupled weight decay regularization," in Proc. Int. Conf. Learn. Represent. (ICLR), New Orleans, LA, USA, 2019. [Online]. Available: https://arxiv.org/abs/1711.05101
[17] C. Y. Wang, I. H. Yeh, and H. Y. M. Liao, "YOLOv9: learning what you want to learn using programmable gradient information," arXiv preprint arXiv:2402.13616, Feb. 2024. [Online]. Available: https://arxiv.org/abs/2402.13616
[18] R. Kute, A. Vyas, and A. Bhatt, "Comparative analysis of YOLOv8 and YOLOv9 models for real-time plant disease detection in hydroponics," Eng. Technol. Appl. Sci. Res., vol. 14, no. 5, pp. 16399–16405, Sep. 2024, doi: 10.48084/etasr.8301.
[19] T. Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan, and S. Belongie, "Feature pyramid networks for object detection," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Honolulu, HI, USA, 2017, pp. 2117–2125, doi: 10.1109/CVPR.2017.106.
[20] S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia, "Path aggregation network for instance segmentation," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Salt Lake City, UT, USA, 2018, pp. 8759–8768, doi: 10.1109/CVPR.2018.00913.
[21] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Las Vegas, NV, USA, 2016, pp. 770–778, doi: 10.1109/CVPR.2016.90.
[22] M. Tan and Q. V. Le, "EfficientNet: rethinking model scaling for convolutional neural networks," in Proc. 36th Int. Conf. Mach. Learn. (ICML), Long Beach, CA, USA, 2019, pp. 6105–6114. [Online]. Available: https://arxiv.org/abs/1905.11946
[23] Z. Liu, H. Mao, C. Y. Wu, C. Feichtenhofer, T. Darrell, and S. Xie, "A ConvNet for the 2020s," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), New Orleans, LA, USA, 2022, pp. 11976–11986, doi: 10.1109/CVPR52688.2022.01167.
[24] D. Singh, N. Jain, P. Jain, P. Kayal, S. Kumawat, and N. Batra, "PlantDoc: a dataset for visual plant disease detection," in Proc. 7th ACM IKDD CoDS 25th COMAD, Hyderabad, India, 2020, pp. 249–253, doi: 10.1145/3371158.3371196.
[25] R. Zhao, Y. Zhu, Y. Li, Z. Wang, and Q. Hu, "Domain randomization meets data augmentation: a unified framework for robust plant disease detection," IEEE Trans. Comput. Imaging, vol. 10, pp. 441–455, 2024, doi: 10.1109/TCI.2024.3362019.
[26] H. Islam and K. Rahman, "PlantVillage revisited: analysis of dataset biases and mitigation strategies for generalization to field conditions," Data Brief, vol. 47, p. 109023, Apr. 2023, doi: 10.1016/j.dib.2023.109023.
[27] S. T. Al-Janabi, M. Rashid, and A. Hamid, "Hybrid CNN-ViT model for disease detection in rice under uncontrolled outdoor conditions," Agronomy, vol. 13, no. 9, p. 2376, Sep. 2023, doi: 10.3390/agronomy13092376.
[28] W. Dong, Y. Zhang, and X. Li, "Class activation mapping enhancements for interpretable plant pathology detection using gradient-weighted localization," Comput. Electron. Agric., vol. 214, p. 108287, Nov. 2023, doi: 10.1016/j.compag.2023.108287.
[29] T. R. Hughes and M. Salathé, "An open access repository of images on plant health to enable the development of mobile disease diagnostics," PeerJ Comput. Sci., vol. 9, e1483, 2023, doi: 10.7717/peerj-cs.1483.
[30] M. H. Tunio, J. Li, X. Zeng, A. Ahmed, S. A. Shah, H.-U. Shaikh, G. A. Mallah, and I. A. Yahya, "Advancing plant disease classification: a robust and generalized approach with transformer-fused convolution and Wasserstein domain adaptation," Comput. Electron. Agric., vol. 227, p. 109574, 2024, doi: 10.1016/j.compag.2024.109574.
[31] S. A. M. Felicita and B. R. Kavitha, “Enhancing multiclass plant disease classification using GAN-boosted vision transformer with XAI insights,” Frontiers in Plant Science, vol. 16, p. 1649399, 2026, doi: 10.3389/fpls.2025.1649399.