George Paschalidis

Intelligent systems need to understand #3D Objects from single images. However, recent SotA methods work well for unoccluded views & common poses, and prioritize shape over pose. We tackle this by developing #SDFit #ICCV2025 a novel Render-n-Compare framework for 3D objects that: 👉 Is robust to occlusion & uncommon poses 👉 Treats both Shape & Pose as 1st class citizens (1/6🧵)

Why does 3D human-object reconstruction fail in the wild or get limited to a few object classes? A key missing piece is accurate 3D contact. InteractVLM (#CVPR2025) uses foundational models to infer contact on humans & objects, improving reconstruction from a single image. (1/10)

🚀 Introducing InterDyn — our newly accepted CVPR work that explores controllable synthesis of interactive dynamics! Building upon powerful video diffusion models, InterDyn infers future motion and interactions directly from an input image and a dynamic control signal (e.g., a moving hand mask). Check out how we push the boundaries of intuitive physics with video generative models. Project page: interdyn.is.tue.mpg.de Arxiv: arxiv.org/abs/2412.11785 #GenAI #AIGC #VideoGen #ML #ComputerVision #CVPR2025 🧵1/6