VGGT-Omega

GenCeption, a new computer vision model leveraging text-to-video generation, achieves state-of-the-art performance across multiple tasks with significantly less training data than specialized models, …

Researchers introduced GenCeption, a model that uses a pre-trained video generative diffusion backbone as a general-purpose vision learner, achieving state-of-the-art performance on tasks like depth e…