🎓 Currently pursuing a PhD inVisual Perception and Reasoning.

🔍 My research interests revolve aroundVision-Language Models 🧠,Embodied Agents 🤖, andScene Graph Generation 🕸. I am passionate about creating generalist AI models capable of understanding and interacting with complex visual data.

• Visual Generalist Models: Developing models that process diverse visual data (e.g., images, videos, 3D, audio, IMU) to tackle various tasks in perception, reasoning, generation, robotics, and gaming. Notable projects includeEgoLife,Octopus,FunQA, andOtter.

• AI Safety for Foundation Models: Investigating how to mitigate hallucinations in large language models (LLMs) and multimodal models (LMMs). A key contribution is the introduction ofUPD to withhold answers when faced with unsolvable questions.

• PSG Series (2022-2023): Led the development of thePSG,PVSG, andPSG4D models, focusing on relation modeling for scene understanding. I also collaborated on works likeRelate-Anything andPairNet.

• OOD Detection (2021-2022): Led a comprehensivesurvey and developedOpenOOD, a popular codebase for Out-of-Distribution detection in AI safety.

• Prompt Tuning (2022): Contributed to foundational works likeCoOp andCoCoOp for prompt tuning in vision-language models.

• Email:yangjingkang001@gmail.com
• LinkedIn:Jingkang Yang
• Twitter:@JingkangY

Feel free to reach out for collaboration or just to chat about AI and technology!

Thanks for visiting my profile!

PinnedLoading

1. EvolvingLMMs-Lab/EgoLifeEvolvingLMMs-Lab/EgoLifePublic

   [CVPR 2025] EgoLife: Towards Egocentric Life Assistant

   Python 227 14

2. dongyh20/Octopusdongyh20/OctopusPublic

   [ECCV2024] 🐙Octopus, an embodied vision-language model trained with RLEF, emerging superior in embodied visual planning and programming.

   Python 285 19

3. OpenPSGOpenPSGPublic

   Benchmarking Panoptic Scene Graph Generation (PSG), ECCV'22

   Python 441 69

4. OpenOODOpenOODPublic

   Benchmarking Generalized Out-of-Distribution Detection

   Python 937 127

5. EvolvingLMMs-Lab/RelateAnythingEvolvingLMMs-Lab/RelateAnythingPublic

   Relate Anything Model is capable of taking an image as input and utilizing SAM to identify the corresponding mask within the image.

   Python 452 21

6. KaiyangZhou/CoOpKaiyangZhou/CoOpPublic

   Prompt Learning for Vision-Language Models (IJCV'22, CVPR'22)

   Python 1.9k 211