Research map / Vision-language and multimodal / Multimodal LLMs
AgroBench: Vision-Language Model Benchmark in Agriculture
Risa Shinoda, Nakamasa Inoue, Hirokatsu Kataoka, Masaki Onishi, Yoshitaka Ushiku
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| remove image input (text-only) for GPT-4o (ablation) | GPT-4o · AgroBench overall | overall accuracy (%): 73.45 -> 29.71 | Hurt |
| remove image input for disease identification (ablation) | GPT-4o · AgroBench DID | accuracy (%): 64.18 -> 1.93 | Hurt |
| remove image input for disease management (ablation) | GPT-4o · AgroBench DMN | accuracy (%): 87.35 -> 72.58 | No clear effect |
| add one-shot chain-of-thought examples (ablation) | GPT-4 on 100 QAs per task · AgroBench | accuracy | No clear effect |
| add chain-of-thought examples for PID, WID, CMN, TM (ablation) | GPT-4 · AgroBench PID/WID/CMN/TM | accuracy | Improved |
| increase CoT examples from one to three (ablation) | GPT-4 · AgroBench | accuracy | No clear effect |
| scale up model size QwenVLM 7B to 72B | QwenVLM · AgroBench overall | overall accuracy (%): 64.15 -> 70.66 | Improved |
| scale up model size QwenVLM 7B to 72B on weed identification | QwenVLM · AgroBench WID | accuracy (%): 33.17 -> 34.48 | No clear effect |
| compare identification vs management tasks | all VLMs · AgroBench | accuracy: GPT-4o DID 64.18 vs DMN 87.35 | Mixed |
| scale up model size LLaVA-Next 8B to 72B | LLaVA-Next · AgroBench overall | overall accuracy (%): 51.28 -> 59.31 | Improved |
Closest papers by meaning
- AgMMU: A Comprehensive Agricultural Multimodal Understanding Benchmark (NeurIPS 2025 · similarity 0.91)
- Empowering and Assessing the Utility of Large Language Models in Crop Science (NeurIPS 2024 · similarity 0.87)
- AgroOmni: A Large-Scale Multi-view Agricultural Dataset for Cross-Scale Multimodal Reasoning (NeurIPS 2026 · similarity 0.87)
- Can Large Multimodal Models Understand Agricultural Scenes? Benchmarking with AgroMind (NeurIPS 2025 · similarity 0.86)
- Utilizing Vision-Language Models for Detection of Leaf-Based Diseases in Tomatoes (AAAI 2025 · similarity 0.86)
- MIRAGE: A Benchmark for Multimodal Information-Seeking and Reasoning in Agricultural Expert-Guided Conversations (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
