Projects sharing features with Depth Anything 3

This project is an RGB-D image inpainting tool and framework for 3D photo reconstruction. It transforms single 2D images into 3D content by estimating monocular depth and synthesizing missing color and depth data to fill occluded regions. The system uses a layered depth image representation to manage scene boundaries and pixel connectivity. This allows for novel view synthesis, enabling the generation of videos that simulate motion parallax effects from different camera perspectives. The project covers a range of spatial modeling capabilities, including depth map estimation, disparity-based

Meshroom is a node-based photogrammetry software designed to transform collections of two-dimensional images into three-dimensional models and scene geometry. It provides a visual interface for constructing and managing modular data pipelines, allowing users to automate complex computer vision tasks such as feature extraction, depth map estimation, and mesh generation. The software distinguishes itself through a distributed computational framework that dispatches resource-intensive tasks across local hardware or remote render farms. By utilizing a directed acyclic graph execution model, it en

Depth-Anything-V2 is a computer vision foundation model designed for general-purpose spatial understanding and depth perception. It functions as a monocular depth estimation model that predicts relative and absolute depth maps from single images or video sequences. The project provides specialized tools for both relative depth estimation and metric depth calculation, allowing for the determination of absolute physical distances in indoor and outdoor environments. It includes a video depth estimation framework that ensures temporal consistency across sequential frames to maintain stable depth

MiDaS is a PyTorch computer vision library and monocular depth estimation model designed to predict scene depth from single images. It functions as a scene depth predictor that computes distance maps to determine object proximity to the camera. The project enables zero-shot depth transfer, allowing the model to be applied to new datasets or environments without additional training data. It focuses on relative depth regression to predict scale-invariant depth maps. The library includes a real-time depth visualizer for capturing live camera feeds and displaying corresponding depth maps. It als

Depth-Anything is a monocular depth estimation foundation model that produces dense per-pixel depth maps from a single RGB image. It is built on a DINOv2 Vision Transformer encoder backbone and trained on 62 million unlabeled images using a teacher-student pseudo-labeling framework, enabling robust generalization across diverse scenes without task-specific training. The model outputs both relative depth maps, which capture the ordering of scene points, and metric depth maps with real-world units after fine-tuning on datasets like NYUv2 or KITTI. The project distinguishes itself through its ab

ComfyUI-3D-Pack is a suite of custom nodes for ComfyUI that enables 3D asset generation and rendering within a node-based workflow. It provides a set of tools for reconstructing textured three-dimensional meshes and volumetric scenes from single images, multi-view images, or text prompts. The system includes a Gaussian splatting generator for creating high-fidelity volumetric 3D scene representations and a multi-view image generator to produce consistent image sets for reconstruction. It also features a single image 3D mesh tool to build geometry from a single 2D source. The toolset covers 3

Gaussian Splatting is a computational framework designed to transform sparse sets of two-dimensional photographs into photorealistic, interactive three-dimensional scene representations. The system functions as a reconstruction tool and rendering engine, enabling the conversion of image data into volumetric models that support novel view synthesis. The project represents scenes as a collection of anisotropic three-dimensional Gaussians, which store position, opacity, color, and covariance data. It distinguishes itself through a differentiable tile-based rasterization process that projects the

Kornia is a differentiable computer vision library and cross-framework tensor vision toolset. It implements vision operations as differentiable tensors to enable integration into deep learning pipelines and supports the transpilation of operations across PyTorch, TensorFlow, JAX, and NumPy. The project provides specialized toolsets for geometric vision and stereo depth, including algorithms for 3D scene reconstruction, camera calibration, and pose estimation. It further distinguishes itself as a differentiable image augmentation framework, applying random geometric and color transformations w

OpenSfM is a computer vision library and structure-from-motion pipeline designed to reconstruct three-dimensional scenes and camera trajectories from overlapping images. It functions as a 3D reconstruction engine and photogrammetry toolkit, utilizing automated feature-based image matching and incremental bundle adjustment to derive spatial geometry. The system distinguishes itself as a geospatial alignment tool, integrating GPS and inertial sensor data to align reconstructed 3D models with real-world geographic coordinates. It employs a hybrid Python and C++ execution model to manage large-sc

MMSegmentation is an open-source semantic segmentation toolbox built on PyTorch that provides a modular, configurable framework for building, training, evaluating, and deploying segmentation models. At its core, it offers a config-driven pipeline that assembles training, evaluation, and inference workflows by parsing hierarchical configuration files, with a modular component registry that enables plug-and-play composition of neural network modules, optimizers, datasets, and metrics. The framework supports the full model lifecycle through a unified runner interface that controls training, testi

pyslam is a framework for Simultaneous Localization and Mapping that combines Python flexibility with C++ performance. It is a sparse SLAM implementation designed to map environment geometry and track device location by processing image frames into 3D points. The project features a bridge for exposing high-performance C++ classes to Python scripts using zero-copy memory sharing. This integration allows for switching between a scripting interface for rapid prototyping and a compiled core for execution speed. The system includes a spatial map optimizer to refine 3D point and camera pose estima

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

PyTorch3D is a 3D geometric deep learning library and mesh processing toolkit designed for learning from point clouds and complex 3D surface geometries. It provides a collection of reusable components and data structures for deep learning with 3D data, including a framework for training and evaluating neural radiance fields to enable photorealistic view synthesis. The project features a differentiable 3D renderer that converts meshes and point clouds into 2D images while allowing gradients to flow back into the geometry and textures. This enables 3D shape optimization, where mesh geometry, te

This project is a computer vision pipeline and volumetric rendering system used to transform photos and videos into high-fidelity 3D models. It implements a deformable neural radiance field framework that optimizes deformation fields to represent non-rigid moving subjects in three dimensions. The system utilizes volumetric deformation fields to map 3D coordinates from a static canonical space to a deformed state. This allows for the reconstruction of photorealistic scenes and the synthesis of high-fidelity images from camera perspectives not present in the original input data. The framework

openMVS is a multi-view stereo library and photogrammetry pipeline used for 3D scene reconstruction. It transforms Structure from Motion data—specifically camera poses and sparse point clouds—into detailed 3D models consisting of dense point clouds and textured meshes. The project provides a sequence of processing stages to densify point clouds, generate 3D surface meshes, and apply photorealistic textures. It uses multi-view texture blending to map accurate colors onto reconstructed geometry and employs iterative refinement to optimize mesh details. The system includes capabilities for impo

4DGaussians is a research library and neural rendering engine designed for reconstructing and rendering dynamic three-dimensional scenes. It represents moving environments as a collection of Gaussian primitives that evolve in position and appearance over a temporal dimension. The framework utilizes neural deformation fields to predict spatial offsets and rotations for static point representations, simulating complex motion over time. It further employs temporal basis decomposition to encode motion trajectories into learned functions, compressing dynamic scene data while maintaining smooth tra

This project is a computer vision system for monocular depth estimation and 3D point cloud generation. It provides a supervised depth learning framework and a depth predictor capable of estimating spatial distance and disparity from single 2D images using pretrained neural networks. The system includes tools to transform 2D depth images into 3D point clouds via pixel coordinate backprojection and converts 3D point cloud data into 2D depth maps. It utilizes a training pipeline that supports model fine-tuning and hyperparameter optimization. The library covers broader capabilities in spatial a

This project is a monocular depth estimation model and computer vision framework designed to calculate absolute distance and scale from single images. It functions as a metric depth estimator that generates high-resolution depth maps without requiring camera-specific focal length metadata. The system utilizes a vision transformer architecture for feature extraction and zero-shot inference to produce metric-scale depth predictions. It includes specialized components for sharp-boundary depth refinement to maintain high-frequency edge details and prevent blurriness at object boundaries. The rep

The Intel RealSense SDK is a software development kit providing drivers and libraries for interfacing with depth cameras to capture color, depth, and infrared data streams. It includes a depth camera driver for device discovery and sensor configuration, a stereo vision library for computing depth maps and aligning frames, and a 3D point cloud generator to transform depth and infrared frames into spatial representations. The SDK distinguishes itself through on-chip depth calculation and stereo calibration, using internal vision processors to reduce host CPU load. It supports hardware-level str

VideoPose3D is a machine learning framework designed for 3D human pose estimation. It functions as a motion reconstruction tool that predicts 3D joint positions from 2D video sequences using a temporal convolutional network to process body movement over time. The project includes a semi-supervised learning pipeline that improves pose accuracy by combining labeled datasets with unlabeled video data and projection consistency loss. It also features a video pose visualizer capable of rendering 3D skeleton reconstructions and 2D keypoints as overlays on original footage. The framework covers the

This project is a diffusion-based AI art generator and animation framework used to create digital images and motion graphics from text prompts. It functions as a system for producing stylized videos and AI art through iterative diffusion sampling and neural network models. The framework distinguishes itself through specialized tools for 3D depth animation, using depth-map transformations to create spatial movement. It also includes neural style transfer capabilities to apply specific artistic looks, such as watercolor or pixel art, and utilizes optical flow frame blending to reduce flickering

DreamGaussian is a generative system and converter designed to create textured three-dimensional models from text or images using Gaussian Splatting. It functions as a pipeline for transforming two-dimensional inputs into high-fidelity 3D assets. The project provides specific workflows for converting 3D Gaussian point clouds into standard textured mesh formats compatible with external 3D software. It supports the generation of textured meshes from single images via volumetric refinement and UV texture optimization, as well as the creation of 3D models from text prompts through intermediate im

DeepLabCut is a deep learning toolkit for markerless 2D and 3D animal pose estimation. It functions as a motion tracking system that identifies anatomical keypoints on animals in video sequences without the need for physical markers. The framework utilizes transfer learning and a library of pre-trained weights to accelerate the training of networks for different species. It supports multi-individual identity tracking to maintain unique identities across video sequences and offers real-time pose detection for live video feeds. The system covers a broad range of computer vision capabilities, i

DINOv2 is a self-supervised vision transformer foundation model designed to generate high-quality visual representations from raw image data. By leveraging large-scale unlabelled datasets, the framework learns to extract robust numerical embeddings that serve as inputs for various machine learning and analysis workflows. The model distinguishes itself through a teacher-student training framework that utilizes centered and sharpened soft probability distributions to align feature maps across multiple image crops. It incorporates a masking strategy that forces the model to reconstruct missing i

This repository serves as a comprehensive collection of reference implementations for the PyTorch machine learning library. It provides practical examples for building, training, and deploying deep learning models, functioning as a toolkit for developers to explore neural network architectures and training workflows. The project distinguishes itself by offering concrete demonstrations of complex machine learning operations, ranging from computer vision tasks like object detection and depth estimation to the training of large-scale transformer models. These examples illustrate how to implement

sam-3d-body is a machine learning framework for 3D human mesh recovery and pose estimation. It utilizes a 3D human mesh recovery model to reconstruct full-body meshes, including the body, hands, and feet, from a single image. The project implements a specialized extension of the Segment Anything Model to guide the extraction and refinement of human body shapes. This integration allows for prompt-guided mesh recovery, where 2D masks and keypoints constrain the inference of 3D pose and shape parameters. The system covers a range of computer vision capabilities, including 3D spatial alignment t