MapAnything is an end-to-end trained Transformer model that can take multiple modalities as input and directly regress the decomposed metric 3D geometric structure of the scene. This model supports more than 12 different 3D reconstruction tasks, including multi-image SfM, multi-view stereo vision, and monocular metric depth estimation.

Computer VisionSafetensorsEnglish

STream3R is a scalable sequential 3D reconstruction model based on causal Transformer, which redefines point cloud map prediction as a decoder-only Transformer problem. It introduces a streaming processing framework, efficiently processes image sequences using causal attention, and can generalize well to various challenging scenarios, including dynamic scenes where traditional methods often fail.

Computer VisionSafetensors

AnySplat is an advanced 3D Gaussian Splat rendering model that can efficiently generate high-quality 3D scenes from images taken from different perspectives. This model has fast inference capabilities and good generalization performance, providing an innovative solution for 3D reconstruction and rendering.

Computer VisionSafetensors

AerialMegaDepth is a deep learning model focused on aerial-ground reconstruction and view synthesis, capable of reconstructing 3D scenes from aerial images and generating new viewpoints.

Computer VisionSafetensors

LHM is a feedforward model that can reconstruct an animatable 3D human body from a single image within seconds. Trained with image reconstruction loss on a large-scale video dataset, our model demonstrates strong generalization capabilities across diverse real-world scenarios.

dust3r is a deep learning model for generating 3D models from images, supporting multi-view 3D reconstruction.

Computer VisionSafetensors

A family of medical image processing models consisting of six large-scale, generalizable 2D/3D variational autoencoders capable of encoding medical images into compressed latent representations and achieving high-fidelity image reconstruction.

DAS3R is a 3D model designed for static scene reconstruction, utilizing dynamic-aware Gaussian splatting technology to generate high-quality 3D reconstructions from images.

Computer VisionSafetensors

SPAR3D is a large-scale reconstruction model based on SF3D, capable of generating textured UV-unwrapped 3D mesh assets from a single image within 1 second, with real-time editing capabilities.

Computer VisionSafetensors

SLAM3R is a real-time dense scene reconstruction system based on monocular RGB video, achieving 3D scene reconstruction through feedforward neural network point cloud regression.

Computer VisionSafetensors

SLAM3R is a real-time dense scene reconstruction system based on monocular RGB video, capable of performing point cloud regression through feedforward neural networks to achieve dense 3D reconstruction.

Computer VisionSafetensors

VFusion3D is a large-scale feed-forward 3D generation model trained with limited 3D data and extensive synthetic multi-view data, representing the first work to explore scalable 3D generation/reconstruction models.

Computer VisionTransformers

VFusion3D is a large-scale feed-forward 3D generation model trained with limited 3D data and extensive synthetic multi-view data, exploring scalable 3D generation/reconstruction models.

Computer VisionTransformers

TripoSR is a fast feed-forward 3D generation model jointly developed by Stability AI and Tripo AI, specializing in rapid 3D reconstruction from a single image.