At OmAI Lab, we develop multimodal foundation models that connect visual perception, reasoning, and action.

Our research advances visual reasoning and embodied intelligence, with a focus on open-vocabulary perception, reinforcement learning for vision-language models, and real-time inference architectures. We aim to build models that understand dynamic visual environments and translate high-level goals into grounded decisions and actions.

• 📹 VLX-Flow: A streaming model for efficient visual decision-making.
• 🔍 VLX-Seek: Fine-grained visual perception and grounding for physical AI.
• 🚗 VLX-Go: Efficient general-purpose embodied navigation in the wild.
• 🤔 VLX-VR: An agentic video reasoning model for videos of any length.

Models that think, reason, and understand the visual world at a granular level.

• 🌟 VLM-R1: Solving Visual Understanding with Reinforced VLMs. (Highly active)
• 🔎 ZoomEye: Enhancing Multimodal LLMs with human-like zooming capabilities through tree-based image exploration.
• 🌍 ImageRAG: Enhancing ultrahigh-resolution remote sensing imagery analysis.

Foundational spatial understanding optimized for edge and on-premise speeds.

• ⚡ OmDet: Real-time, highly accurate, open-vocabulary end-to-end object detection.
• 🔍 VLM-FO1: Bridging the gap between high-level reasoning and fine-grained perception in Vision-Language Models.
• 📐 GroundVLP: Harnessing Zero-shot Visual Grounding from Vision-Language Pre-training.

Action-oriented intelligence for physical and virtual environments.

• 🛠️ OmAgent: A comprehensive framework to build multimodal language agents for fast prototyping and production.
• 🎯 OmTrackVLA: Open and reproducible research for tracking Vision-Language-Action (VLA) models.

Rigorous standards for the open-source multimodal community.

• 🌍 RS5M: A pioneer work in VLM benchmark for remote sensing.
• 📏 OVDEval: A comprehensive evaluation benchmark for Open-Vocabulary Detection.
• 📝 VL-CheckList: Evaluating vision & language pretraining models with objects, attributes, and relations.