WeVisOne-VL Unifies Object and Pixel-Level Perception in One Open VLM
The paper presents WeVisOne-VL, a vision-language model that unifies object-level and pixel-level perception through structured next-token prediction. It supports object detection, referring localization, counting, OCR, segmentation, and matting, integrating a lightweight pixel decoder via a special <OBJ> token. Training uses a three-stage curriculum SFT followed by mixed-task RL. The authors report leading performance across over 30 benchmarks and robust rejection of object-absent queries, and state the model is released to the open-source community.
One primary source; other links are context, not independent corroboration.
- Primary sourceX discovery · automated brief
- Discovery linkX
- Linked in sourceView PDF
- Linked in sourcearXiv:2610.31329v1
- Linked in sourceLearn more about arXivLabs
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
Open full article view →- Original paper+45
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
The paper presents WeVisOne-VL, a vision-language model that unifies object-level and pixel-level perception through structured next-token prediction. It supports object detection, referring localization, counting, OCR, segmentation, and matting, integrating a lightweight pixel decoder via a special <OBJ> token. Training uses a three-stage curriculum SFT followed by mixed-task RL. The authors report leading performance across over 30 benchmarks and robust rejection of object-absent queries, and state the model is released to the open-source community.
Key findings from the stored brief
- Unifies object-level and pixel-level perception through structured next-token prediction in a single VLM.
- Integrates a lightweight pixel decoder via a special <OBJ> token to produce instance-level pixel outputs.
- Trained with a three-stage curriculum SFT followed by mixed-task RL.
- Reports leading performance across over 30 benchmarks while rejecting object-absent queries.
Source claims not independently checked; benchmark results are reported by the authors.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
These are Grok Bot discovery notes. AINewsIQ checked the primary page for its shared brief; the bot's rationale, linked posts, authors, and stances have not been independently verified.
Tencent's WeChat Vision team presents WeVisOne-VL, a VLM that adds a lightweight pixel decoder via an <OBJ> token so one model handles detection, referring localization, counting, OCR, segmentation and alpha matting, and reports leading results across 30+ benchmarks (self-reported). 2B and 4B weights are on Hugging Face; a hands-on 8GB test found it worked but hallucinated objects.
Source claims not independently checked; benchmark results are reported by the authors.
- Bot tagged: questions Running it on an 8GB RTX 4060 laptop, it counted crowds, cut out a cat and read a neon sign, but hallucinated giraffes. Inspect post ↗
Engagement counts, when shown, are observed snapshots. They do not establish accuracy or corroboration.
Related archive coverage
These relationships record a citation or topic overlap, not confirmation of the finding.
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
Combining detection, OCR, segmentation and matting in one open model could make fine-grained visual perception more reliable for real-world deployment, and its open release lets the community test these claims directly.
Limits and open questions
Source claims not independently checked; benchmark results are reported by the authors.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Ask about the source, add context, or point readers to a relevant follow-up.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.