AgroBench: Vision-Language Model Benchmark in Agriculture

TL;DR

AgroBench is a comprehensive, expert-annotated benchmark for evaluating vision-language models in agriculture, highlighting current limitations and guiding future improvements in fine-grained agricultural task understanding.

Contribution

Introduces AgroBench, a new expert-annotated benchmark covering diverse agricultural categories to evaluate and analyze vision-language models in real-world farming tasks.

Findings

VLMs need improvement in fine-grained identification tasks.

Most open-source VLMs perform near random in weed identification.

The benchmark reveals specific error patterns and development pathways.

Abstract

Precise automated understanding of agricultural tasks such as disease identification is essential for sustainable crop production. Recent advances in vision-language models (VLMs) are expected to further expand the range of agricultural tasks by facilitating human-model interaction through easy, text-based communication. Here, we introduce AgroBench (Agronomist AI Benchmark), a benchmark for evaluating VLM models across seven agricultural topics, covering key areas in agricultural engineering and relevant to real-world farming. Unlike recent agricultural VLM benchmarks, AgroBench is annotated by expert agronomists. Our AgroBench covers a state-of-the-art range of categories, including 203 crop categories and 682 disease categories, to thoroughly evaluate VLM capabilities. In our evaluation on AgroBench, we reveal that VLMs have room for improvement in fine-grained identification tasks. Notably, in weed identification, most open-source VLMs perform close to random. With our wide range of topics and expert-annotated categories, we analyze the types of errors made by VLMs and suggest potential pathways for future VLM development. Our dataset and code are available at https://dahlian00.github.io/AgroBenchPage/ .