---
title: VLM-FO1-3B-v01
canonical_url: "https://www.modelscope.cn/models/Om_AI_Lab/VLM-FO1-3B-v01"
md_url: "https://www.modelscope.cn/models/Om_AI_Lab/VLM-FO1-3B-v01.md"
repository: Om_AI_Lab/VLM-FO1-3B-v01
last_updated: 2026-06-18
license: apache-2.0
model_type:
  - omchat_qwen2_5_vl
architectures:
  - OmChatQwen25VLForCausalLM
library_name:
  - safetensors
  - pytorch
frameworks:
  - Pytorch
language:
  - zh
  - en
downloads: 167
stars: 2
tags:
  - object-detection
  - multimodal
  - REC
  - VLM
  - zero-shot-object-detection
---

# VLM-FO1-3B-v01

> VLM-FO1-3B-v01 - Om_AI_Lab 在 ModelScope 开源的模型。VLM-FO1: VLM-FO1-3B-v01

Om_AI_Lab/VLM-FO1-3B-v01 是 ModelScope 魔搭社区上的机器学习模型，采用 apache-2.0 许可。

- **Repository**: Om_AI_Lab/VLM-FO1-3B-v01
- **License**: apache-2.0
- **Tags**: object-detection, multimodal, REC, VLM, zero-shot-object-detection
- **Downloads**: 167
- **Stars**: 2
- **Last updated**: 2026-06-18

Source: https://www.modelscope.cn/models/Om_AI_Lab/VLM-FO1-3B-v01

---

# VLM-FO1: VLM-FO1-3B-v01

This repository contains the VLM-FO1-3B-v01 model, an implementation of the [VLM-FO1](https://github.com/om-ai-lab/VLM-FO1) framework.

VLM-FO1 is a novel plug-and-play framework designed to bridge the gap between the high-level reasoning of Vision-Language Models (VLMs) and the need for fine-grained visual perception.

## Model Details

### Model Description

VLM-FO1 endows pre-trained VLMs with superior fine-grained perception without compromising their inherent high-level reasoning and general understanding capabilities. It operates as a plug-and-play module that can be integrated with any existing VLM, establishing an effective and flexible paradigm for building the next generation of perception-aware models.

VLM-FO1 excels at a wide range of fine-grained perception tasks, including Object Grounding, Region Generative Understanding, Visual Region Reasoning, and more.

🧩 **Plug-and-Play Modularity:** Our framework is designed as a set of enhancement modules that can be seamlessly integrated with any pre-trained VLM, preserving its original weights and capabilities.

🧠 **Hybrid Region Encoder (HFRE):** We introduce a novel Dual-Vision Encoder architecture that fuses semantic-rich features with perception-enhanced features, creating powerful region tokens that capture both high-level meaning and fine-grained spatial detail.

🎯 **State-of-the-Art Performance:** VLM-FO1 achieves SOTA results across a diverse suite of benchmarks.

✅ **Preserves General Abilities:** Our two-stage training strategy ensures that fine-grained perception is gained without causing catastrophic forgetting of the base model's powerful general visual understanding abilities.

### Model Sources

- **Repository:** [https://github.com/om-ai-lab/VLM-FO1]
- **Paper:** [https://arxiv.org/pdf/2509.25916]



## Citation

```bibtex
@article{liu2025vlm,
  title={VLM-FO1: Bridging the Gap Between High-Level Reasoning and Fine-Grained Perception in VLMs},
  author={Liu, Peng and Shen, Haozhan and Fang, Chunxin and Sun, Zhicheng and Liao, Jiajia and Zhao, Tiancheng},
  journal={arXiv preprint arXiv:2509.25916},
  year={2025}
}
```