MLLMs adapted for electron microscopy segmentation prompts

Researchers have explored the use of open-weight multimodal large language models (MLLMs) to generate point prompts for electron microscopy segmentation. By fine-tuning models like Qwen3-VL with LoRA adapters on existing datasets, they achieved significant improvements in segmentation accuracy, with Qwen3-VL reaching an AP50 score of 0.736. While this approach shows promise in linking language instructions to mask decoding, a supervised centroid-heatmap detector still outperforms it on specific tasks. AI

IMPACT Demonstrates a new method for using MLLMs in scientific image analysis, potentially improving automation in microscopy.

RANK_REASON Academic paper detailing a novel application of MLLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• Automatic Prompt Generation (APG)
• Electron Microscopy Segmentation
• Hugging Face
• LoRA adapters
• Open-Weight MLLMs
• Qwen3 VL
• supervised centroid-heatmap detector

• paper
• model release

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