![[2508.07514] Robust MultiSpecies Agricultural Segmentation Across Devices, Seasons, and Sensors Using Hierarchical DINOv2 Models](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2508.07514] Robust MultiSpecies Agricultural Segmentation Across Devices, Seasons, and Sensors Using Hierarchical DINOv2 Models
Summary
This article presents a robust segmentation framework using Hierarchical DINOv2 models for reliable plant species and damage identification in agricultural settings, addressing challenges across diverse conditions.
Why It Matters
The study highlights the need for advanced machine learning models that can generalize across varying agricultural environments. By improving segmentation accuracy, this research can enhance operational phenotyping workflows, ultimately benefiting agricultural productivity and sustainability.
Key Takeaways
- The DINOv2 model significantly improves species-level segmentation accuracy under varying conditions.
- Hierarchical taxonomic inference enhances robustness against environmental shifts.
- Error analysis identifies key challenges, such as vegetation-soil confusion, in segmentation tasks.
- The framework is successfully integrated into BASF's phenotyping workflow, demonstrating practical applications.
- Training on a diverse dataset from multiple years and locations strengthens model generalizability.
Computer Science > Computer Vision and Pattern Recognition arXiv:2508.07514 (cs) [Submitted on 11 Aug 2025 (v1), last revised 16 Feb 2026 (this version, v2)] Title:Robust MultiSpecies Agricultural Segmentation Across Devices, Seasons, and Sensors Using Hierarchical DINOv2 Models Authors:Artzai Picon, Itziar Eguskiza, Daniel Mugica, Javier Romero, Carlos Javier Jimenez, Eric White, Gabriel Do-Lago-Junqueira, Christian Klukas, Ramon Navarra-Mestre View a PDF of the paper titled Robust MultiSpecies Agricultural Segmentation Across Devices, Seasons, and Sensors Using Hierarchical DINOv2 Models, by Artzai Picon and 8 other authors View PDF HTML (experimental) Abstract:Reliable plant species and damage segmentation for herbicide field research trials requires models that can withstand substantial real-world variation across seasons, geographies, devices, and sensing modalities. Most deep learning approaches trained on controlled datasets fail to generalize under these domain shifts, limiting their suitability for operational phenotyping pipelines. This study evaluates a segmentation framework that integrates vision foundation models (DINOv2) with hierarchical taxonomic inference to improve robustness across heterogeneous agricultural conditions. We train on a large, multi-year dataset collected in Germany and Spain (2018-2020), comprising 14 plant species and 4 herbicide damage classes, and assess generalization under increasingly challenging shifts: temporal and device changes ...