Moondream releases RefCOCO-M with pixel-accurate masks and cleaned prompts for grounding evals
victormustar · x · 2026-10-31
Moondream announced RefCOCO-M, a refreshed version of the classic referring expression segmentation benchmark RefCOCO:
- Ships with pixel-accurate masks replacing the old coarse annotations — comparison images show visibly tighter segmentation boundaries
- Problematic prompts removed, cutting ambiguity and low-quality samples that skew evaluations
- Open-sourced on Hugging Face (moondream/refcoco-m) under MIT license, citing the original arXiv:1603.06180 paper
The goal is better data for more reliable evaluation of grounding and referring segmentation models.
More from Research
- NVIDIA Nemotron 3.5 scales LLM self-verification, +9-10 points on agentic benchmarks — sarahcat21 · 2026-10-31
- Unofficial OpenTargets MCP Server Brings Gene-Drug-Disease Data to LLMs — modelcontextprotocol · 2026-10-31
- Researchers find 5 personality structures in LLM weight space, Big Five-like — luke_drago_ · 2026-10-31
- NVS benchmarks are misaligned: proper pose alignment puts 3D Gaussians back at SOTA — kwangmoo_yi · 2026-10-31
- arXiv moderator: submission surge causes months-long holds, new moderators wanted — tdietterich · 2026-10-31
- arXiv policy checks reject class projects, proposals, surveys and position papers — tdietterich · 2026-10-31