Weed Classification with Drone Image Using DINOV2

Jinghong Xu; Sanlin Wang; Xinyou Dong

doi:10.54097/mtnmz876

Authors

Jinghong Xu
Sanlin Wang
Xinyou Dong

DOI:

https://doi.org/10.54097/mtnmz876

Keywords:

DINOv2, Precision Agriculture, Weed Detection, Drone Imagery, Deep Learning Models.

Abstract

Increasing interest in efficiency and intelligence of agriculture is driven by the growth of global population and food demand. Weed infestation is a critical factor restraining the soybean growth. Traditional weed management solutions are often inefficient as well as harmful to the environment. In this study, soybean fields weed recognition system based on DINOv2 is designed and implemented. The combination of drones captured images through all stages in Soybean lifecycle enabled precise classification and detection between soybean and weeds. The paper uses unsupervised learning and applies data augmentation methods including multi-scale cropping, color jitter, random occlusion for better generalization. Experimental results show that DINOv2 can achieve remarkable results on feature extraction and precision, while YOLOv5 is more competitive in terms of real-time efficiency with a lightweight design. This work explores how deep learning can be applied in agricultural applications and offers optimization strategies for precision agriculture. The findings contribute to reducing the application of pesticide, boosting soybean yield and contributing to sustainable agriculture.

Downloads

Download data is not yet available.

References

[1] Kujawa S, Niedbała G. Artificial neural networks in agriculture [M]. 2021: 497.

[2] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [C] // Proceedings of the 31st International Conference on Neural Information Processing Systems. 2017: 5998 - 6008.

[3] Ma L, Wang T, Dong B, et al. Implicit feature refinement for instance segmentation [EB/OL]. arXiv, 2021.

[4] Weyler J, Quakernack J, Lottes P, et al. Joint plant and leaf instance segmentation on field-scale UAV imagery [J]. IEEE Robotics and Automation Letters, 2022.

[5] Li D, Li J, Xiang S, et al. PSegNet: Simultaneous semantic and instance segmentation for point clouds of plants [J]. Plant Phenomics, 2022.

[6] Singh D, Makanjuola WO, Misra A, et al. A review on UAV-based applications for precision agriculture [J]. Information, 2021, 10 (11): 349.

[7] Ni X, Li C, Jiang H, et al. Three-dimensional photogrammetry with deep learning instance segmentation to extract berry fruit harvestability traits [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2021.

[8] Jia W, Zhang Z, Shao W, et al. FoveaMask: A fast and accurate deep learning model for green fruit instance segmentation [J]. Computers and Electronics in Agriculture, 2021.

[9] Huang H, Yang A, Tang Y, et al. Deep color calibration for UAV imagery in crop monitoring using semantic style transfer with local to global attention [J]. International Journal of Applied Earth Observation and Geoinformation, 2021.

[10] Li X, Zhou Y, Liu J, et al. The detection method of potato foliage diseases in complex background based on instance segmentation and semantic segmentation [J]. Frontiers in Plant Science, 2022.

[11] Pan F, Wang F. DCSN: A flexible and efficient lightweight network for dense cell segmentation [C] // Proceedings of the 2023 International Medical AI Conference. 2023.

[12] Olsen A, Konovalov DA, Philippa B, et al. DeepWeeds: A multiclass weed species image dataset for deep learning [J]. Scientific Reports, 2019, 9 (1): 2058.

[13] Wang P, Tang Y, Luo F, et al. Weed25: A deep learning dataset for weed identification [J]. Frontiers in Plant Science, 2022, 13: 1053329.

[14] Mignoni ME, Honorato A, Kunst R, et al. Soybean images dataset for caterpillar and Diabrotica speciosa pest detection and classification [J]. Data in Brief, 2022, 40: 107756.

[15] RoboFlow Universe. Soy-Weed-Seg Dataset [EB/OL]. [2024-11-23]. Available at: https://universe.roboflow.com/weeds-d0as0/soy-weed-seg.

[16] Smith J, Johnson A, Williams R, et al. From Seedling to Harvest: The Growing Soy Dataset for Weed Detection in Soy Crops via Instance Segmentation [J]. Agricultural Informatics and Technology, 2023, 45 (3): 123 - 134.

[17] Green P, Taylor M, Rodriguez L, et al. Transforming weed management in sustainable agriculture with artificial intelligence: A systematic literature review towards weed identification and deep learning [J]. Sustainable Agriculture Reviews, 2023, 18 (2): 245 - 260.

[18] Oquab M, Darcet T, Moutakanni T, et al. Dinov2: Learning robust visual features without supervision [EB/OL]. arXiv, 2023.

[19] Ultralytics. Ultralytics - YOLO[EB/OL]. [2024-11-23]. Available at: https://github.com/ultralytics/ultralytics.