Projects sharing features with Coma

This project is a diffusion-based 3D generator and image-to-3D reconstruction system. It translates natural language descriptions or two-dimensional images into three-dimensional assets using neural radiance fields and diffusion models. The system utilizes score-distillation sampling and diffusion-based guidance to refine 3D shapes without requiring 3D training data. It includes specialized tools for transforming neural representations into exportable meshes with texture and material data, as well as a pipeline for iterative optimization of geometry and textures. The project covers a broad r

InstantMesh is a neural 3D reconstruction tool and single-image 3D mesh generator. It utilizes a sparse-view large reconstruction model to convert a single two-dimensional image into a three-dimensional object mesh. The system functions as a textured 3D mesh exporter, saving generated objects with either vertex colors or full texture maps for use in external rendering software. The framework covers a range of capabilities including feed-forward geometry inference, single-image depth estimation, and neural radiance fields. It also supports differentiable mesh rendering and workflows for spars

This project is a collection of neural network models and geometric tools designed for image feature matching, spatial alignment, and visual localization. It provides a pre-trained neural network model for identifying high-accuracy correspondences between sparse image features without requiring local training. The system utilizes a graph neural network matcher that employs attention mechanisms and message passing to learn spatial relationships between image feature points. It integrates a RANSAC camera pose estimator to filter feature matches and calculate the relative spatial transformation

This project is a computer vision system for object segmentation and tracking across images and videos. It employs models capable of identifying and masking objects using text prompts, bounding boxes, click points, or image exemplars. The system differentiates itself through memory-based video tracking and shared-memory architectures that maintain consistent object identities over time. It supports multi-object processing in single computation passes to increase frame throughput and utilizes iterative refinement to correct segmentation boundaries through sequential prompts. The software also

jetson-inference is a set of libraries and tools for executing optimized deep learning models on embedded GPU hardware. Its primary purpose is to enable real-time computer vision and AI inference at the edge with low latency and high throughput. The project distinguishes itself through high-performance streaming analytics and the ability to execute concurrent AI pipelines on auto-grade silicon. It provides specialized support for multi-sensor stream processing, utilizing zero-copy data transport to load camera frames directly into GPU memory. The codebase covers a broad surface of capabiliti

Slambook is a visual SLAM framework designed for simultaneous localization and mapping. It provides an integrated system to estimate camera motion and reconstruct 3D environments using visual sensor data. The project includes a visual odometry engine to track camera movement and a dense 3D reconstruction tool for creating volumetric representations of scenes. It features a loop closure detection system to recognize previously visited locations and a pose graph optimizer to refine trajectories and ensure global map consistency. The framework covers spatial estimation and environment modeling

Three-dimensional content creation has been a central research area in computer graphics for decades. The main challenge is to minimize manual intervention, while still allowing the creation of a variety of plausible 3D objects. In this work, we present a global-to-local generative model to…

Project Page

In this work, we propose an efficient iterative planar parameterization for disk topology shapes. The parameterization is used as a tool to regularize the mesh onto a square grid and encoded with vertex position. The resultant encoding is an image with rgb denoting xyz positions on the mesh.

This repository contains source code for Weakly supervised 3D Reconstruction with Adversarial Constraint. This is a fork project of our previous work, 3D-R2N2: 3D Recurrent Reconstruction Neural Network. Inspired by visual hull algorithm, we propose to learn 3D reconstruct from 2D silhouettes…

This repository contains a tensorflow implementation for the paper "Learning Descriptor Networks for 3D Shape Synthesis and Analysis ". (http://www.stat.ucla.edu/~jxie/3DDescriptorNet/3DDescriptorNet.html)

This is the code for NIPS15 paper Weakly-supervised disentangling with recurrent transformations for 3D view synthesis by Jimei Yang, Scott Reed, Ming-Hsuan Yang and Honglak Lee.

By Jun Li, Kai Xu, Siddhartha Chaudhuri, Ersin Yumer, Hao Zhang, Leonidas Guibas

Source code accompanying the ECCV'16 paper "Multi-view 3D Models from Single Images with a Convolutional Network" by M. Tatarchenko, A. Dosovitskiy and T. Brox http://lmb.informatik.uni-freiburg.de/people/tatarchm/mv3d/. The models implemented here are slightly different from those described in…

Source code accompanying the paper "Octree Generating Networks: Efficient Convolutional Architectures for High-resolution 3D Outputs" by M. Tatarchenko, A. Dosovitskiy and T. Brox. The implementation is based on Caffe, and extends the basic framework by providing layers for octree-specific features.

Sketch-Based 3D Exploration with Stacked Generative Adversarial Networks

Arxiv: https://arxiv.org/abs/2012.09235

@inproceedings{firman-cvpr-2016, author = {Michael Firman and Oisin Mac Aodha and Simon Julier and Gabriel J Brostow}, title = {{Structured Completion of Unobserved Voxels from a Single Depth Image}}, booktitle = {Computer Vision and Pattern Recognition (CVPR)}, year = {2016} }

3DMM ¯¯¯¯ This software is an implementation of the 3D morphable model, as defined by Volker Blanz and Thomas Vetter in "A Morphable Model For The Synthesis Of 3D Faces" (SIGGRAPH 99).

We propose a new meshing algorithm to generate a surface with correct topology for the output points of a point network. GAMesh can be used both in post-processing to mesh the output points or to train the point network to directly optimize the vertex positions of the final 3D mesh. Unlike…

We propose a new point-to-surface based loss function named Quadric Loss, which minimizes the quadric error between the reconstructed points and the input surface. Unlike Chamfers or L2 which are spherical losses (equidistant points have equal error), Quadric loss is a ellipsoidal loss, which…

Created by Nadav Schor , Oren Katzir , Hao Zhang , Daniel Cohen-Or .

This is the official inference code for:

This repository contains the TensorFlow implementation for the following paper

[Project](https://ranahanocka.github.io/ALIGNet/) [Arxiv](https://bit.ly/alignet) ALIGNet is a network trained to register pairs of shapes using a learned data-driven prior, and doesn't need ground-truth warp fields for supervision.

This is an official FLAME repository.

Project page of paper "Soft Rasterizer: A Differentiable Renderer for Image-based 3D Reasoning"