New PlantMicro benchmark reveals VLM struggles with microscopic image understanding

A new benchmark called PlantMicro has been developed to evaluate the capabilities of vision-language models (VLMs) in understanding microscopic plant images. The benchmark includes over 5,000 images and 9,000 question-answer pairs designed to test fine-grained recognition and reasoning. Current VLMs, including GPT-5, show significant limitations in this domain, with GPT-5 achieving only 34.93% accuracy on a pathogen classification task, highlighting a gap in their ability to comprehend microscopy-level plant imagery. AI

IMPACT Highlights limitations in current VLMs for specialized scientific domains, potentially guiding future model development for microscopy applications.

RANK_REASON The cluster contains a research paper introducing a new benchmark for evaluating AI models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• DagsHub
• GPT-5
• Hugging Face
• PlantMicro
• Tianqi Wei

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →