Federico Tombari

SG-Bot offers a fresh approach to robotic object rearrangement using observation and planning.

ProText enhances vision-language models using text-only data for better task handling.

This method simplifies adding objects to 3D environments using text and 2D boxes.

FocusCLIP improves human action recognition using heatmaps and text descriptions.

KP-RED improves 3D modeling by using key points for better accuracy and efficiency.

This method simplifies blending different LoRA models for high-quality image generation.

EchoScene enhances indoor 3D scene creation through innovative methods and user interaction.

Assessing the capabilities and challenges of advanced video understanding models.

A new approach enhances mapping and tracking using RGB images.

MiDiffusion improves indoor scene creation using floor plans and object attributes.

Stylebreeder enables users to generate and personalize art with text prompts.

A new diffusion-based approach tackles multiple computer vision tasks effectively.

VQA models may expose private information despite advanced techniques.

New methods improve clarity and accuracy of 3D point clouds.

A new approach enhances the process of filling in incomplete 3D shapes.

G2SDF improves the clarity of 3D images quickly and efficiently.

Learn how SmileSplat creates 3D images from just a few pictures.

Transform static 3D models into lively animations with text commands.

Learn how ESCAPE is revolutionizing shape completion in 3D computer vision.

Learn how new methods create unique images from various themes.

SuperGSeg brings clarity to complex 3D scenes through advanced segmentation techniques.

A new method simplifies image editing without extensive examples.