HuggingFace Daily Papers·· 1 天前
微信推出 WeVisOne-VL：统一目标级与像素级视觉感知
WeVisOne-VL: Unified Visual Perception from Objects to Pixels
AI 导读
微信发布 WeVisOne-VL（WeChat Vision All-in-One）视觉语言模型，通过结构化下一 token 预测统一目标级与像素级感知，支持检测、指代定位、计数、OCR、分割与抠图等任务。模型以特殊 token 接入轻量像素解码器输出实例级像素结果，训练采用三阶段课程式 SFT 加混合任务 RL。在 30 多个基准上，该模型在两类任务上均取得领先表现，并已开源。
正文 · 原文
Abstract:Existing vision-language models (VLMs) remain unreliable for fine-grained visual perception. In this report, we present WeVisOne-VL (WeChat Vision All-in-One), a powerful VLM that unifies object-level and pixel-level perception through structured next-token prediction. It supports multiple visual tasks including object detection, referring localization, counting, OCR, segmentation, and matting, enabling localization at any user-specific granularity, from points and boxes to masks and alpha mattes. Architecturally, we integrate a lightweight pixel decoder into the VLM via a special <OBJ> token to generate instance-level pixel outputs. We train the model with a three-stage curriculum SFT followed by mixed-task RL. Across over 30 benchmarks, WeVisOne-VL demonstrates leading performance on both object-level and pixel-level tasks, while maintaining robust rejection for object-absent queries. By releasing WeVisOne-VL to the open-source community, we hope to advance VLMs from broad visual understanding toward truly reliable perception systems for real-world deployment.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2610.31329 [cs.CV] |
| (or arXiv:2610.31329v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.31329 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tianming Liang [view email]
[v1]
Thu, 22 Oct 2026 14:19:50 UTC (32,841 KB)
来源：HuggingFace Daily Papers · arxiv.org