An open-source, Raspberry Pi-based weed detection system that uses green-detection algorithms to trigger spot spraying for precision weed control.

A toolkit providing standardized, easy-to-use implementations of popular object detection metrics like Average Precision and IOU to ensure consistent benchmarking across different datasets.

The depthai library is the software framework for the Luxonis platform, allowing developers to create AI and computer vision applications on Luxonis hardware.

A medical image segmentation framework featuring MSNet and M2SNet architectures that use multi-scale subtraction to accurately isolate pathologies like polyps and lung infections.

LightM-UNet is a lightweight fusion of UNet and Mamba for medical image segmentation, offering high performance with only 1M parameters.

A free online image annotation tool for generating computer vision ML training data with support for automated bounding box generation and dataset format conversion.

A high-performance inference deployment tool for YOLO-series models on NVIDIA devices, optimizing preprocessing and post-processing using TensorRT and CUDA.

A comprehensive pipeline for Chinese OCR in natural scenes, combining text orientation detection, CTPN for region detection, and CRNN for end-to-end character recognition.

A library for benchmarking and reproducing point-based methods for point cloud understanding, providing a standardized engine for models like PointNet and PointNeXt.

A set of Unity samples demonstrating how to use the Passthrough Camera API to access raw camera data and metadata from Quest 3/3S headsets for MR applications.

CARS is an open-source multiview stereo framework that uses photogrammetry to produce Digital Surface Models (DSM) from satellite imagery.

A self-hosted photo management app that uses object recognition and color analysis to automatically organize and filter photo collections.

A PyTorch implementation of a multi-scale CNN-LSTM model for diagnosing bearing faults in rotating machinery using vibration data.

A flexible and extensible gait analysis framework for human identification and health screening, supporting multiple modalities like silhouettes, skeletons, and LiDAR point clouds.

A patch for Synology Photos that enables facial and object recognition on unsupported NAS hardware by bypassing GPU requirements.

A progressive active learning framework for infrared small target detection that enables models to achieve high performance using only single-point supervision instead of full masks.

An Android video player based on libmpv that uses the Anime4K algorithm to provide real-time super-resolution upscaling for anime and animation videos.

A computer vision-based automation tool that identifies and interacts with screen elements using image recognition to simulate mouse and keyboard input.

An automated structural MRI morphometry system that integrates CAT12 processing to measure brain tissue volumes and generate quantitative reports.

A human-centric framework for assessing the stealthiness of adversarial perturbations in images, featuring a new dataset (APSD) and an attention-based prediction model (A2SM).

An open-source tool providing a semi-automatic OCR workflow specifically designed for high-quality text recognition of historical printings and early printed books.

An open-source framework for building real-time 3D structured light scanners using a single camera and light projector.

An Android QR code recognition library that ports the OpenCV-based WeChat QR engine for faster, multi-code scanning and decoding.

Wis3D is a web-based 3D visualization tool for computer vision researchers that allows for the direct import and viewing of 3D bounding boxes, point clouds, and meshes from Python.

An AI-powered tool that uses Florence-2 and LaMA models to detect and seamlessly remove watermarks from images and AI-generated videos.

LoG is a framework for training and real-time rendering of highly realistic urban-scale 3D models using Gaussian Splatting on a single RTX 4090 GPU.

A detector-free Structure from Motion (SfM) system that reconstructs 3D scenes from images without relying on traditional feature detectors, achieving top performance in the Image Matching Challenge 2023.

An open-source toolbox for markerless human motion capture and novel view synthesis from RGB videos, supporting various input sources from multi-view setups to internet videos.

GVHMR is a SIGGRAPH‑Asia 2024 research codebase that recovers world‑grounded 3‑D human motion from monocular video. It combines visual odometry (DPVO or SimpleVO) with a gravity‑aligned coordinate system, provides ready‑to‑run demo scripts, Colab/HuggingFace demos, and scripts to reproduce benchmark results on 3DPW, RICH, and EMDB.

A PyTorch library for accelerating neural volumetric video research, providing tools for capturing, reconstructing, and rendering volumetric video.

Deep Snake is an open‑source CVPR 2020 implementation for real‑time instance segmentation using a learnable active‑contour (snake) module. The repo ships pre‑trained models for Cityscapes, KITTI and SBD, and provides command‑line tools for evaluation, speed testing, visualisation, demo inference and training (two‑stage for Cityscapes, single‑stage for KITTI/SBD). All functionality is driven by YAML config files; TensorBoard logs are available for monitoring. Suitable for researchers and engineers needing fast, accurate mask predictions.

A unified database and toolkit for human motion capture, consolidating various marker-based datasets into a common framework for animation and deep learning training.

A free, open-source toolbox of Google Colab notebooks with a graphical interface that allows microscopy researchers to train and use deep-learning networks without coding expertise.

OnePose++ is a keypoint-free, one-shot object pose estimation system that determines an object's 3D position and orientation without requiring CAD models.

AutoRecon is a research‑grade pipeline that automatically discovers individual objects in a set of RGB images and reconstructs them as neural SDF surfaces, exporting standard mesh files. It combines a coarse object‑decomposition stage (AutoDecomp) with a neural surface reconstruction stage built on nerfstudio/sdfstudio, and provides ready‑to‑run scripts for demo, BlendedMVS, and CO3D datasets.

MUSt3R is a research‑grade Python library for multi‑view 3‑D reconstruction. It extends the DUSt3R model with symmetric processing, a multi‑layer memory, and online pose estimation, enabling both offline reconstruction of unordered image sets and live SLAM‑style mapping from video or webcam streams. The repo provides installation scripts, several pre‑trained ViT‑based checkpoints (224 px and 512 px), Gradio/viser and Open3D demo interfaces, and a full training pipeline. It is released under a non‑commercial license with restrictive dataset terms.

A PyTorch library that implements the Perceiver and Perceiver IO architectures (and a small experimental variant). It provides ready‑to‑use `Perceiver`, `PerceiverIO`, and `PerceiverLM` classes, supports Fourier positional encoding, and is installable via `pip install perceiver-pytorch`. The repo is a genuine AI/ML project focused on a frontier transformer model for handling large, multimodal inputs.

A lightweight foundation backbone for remote sensing that uses a Large Selective Kernel mechanism to dynamically adjust receptive fields for better object detection and segmentation.

A neural 3D reconstruction framework designed to reconstruct high-quality 3D meshes from images captured in uncontrolled, real-world environments.

PyMAF and PyMAF-X are frameworks for regressing 3D human pose and shape from monocular images and videos using a pyramidal mesh alignment feedback loop.

ENeRF is a research codebase that implements Efficient Neural Radiance Fields, enabling interactive (~20‑50 FPS) free‑viewpoint video. It provides scripts for training on DTU, fine‑tuning on specific scans or human capture data, evaluation with standard metrics, and a GUI for real‑time rendering. The repository includes dataset download instructions, a pretrained DTU model, and detailed commands for multi‑GPU training and evaluation.

A portable C++ library for high-performance image and video segmentation using SAM 2, 2.1, 3, and EdgeTAM, eliminating the need for Python and CUDA.

A text-guided medical image segmentation model that uses medical concepts to identify and segment targets across diverse modalities without needing bounding boxes or points.

An attention-centric real-time object detection framework that combines the accuracy of attention mechanisms with the speed of CNN-based detectors.

OpenSeeFace is an open‑source, CPU‑only facial‑landmark tracker (66 points) built on a MobileNetV3 model exported to ONNX. It streams landmark, gaze, and blink data over UDP for real‑time avatar animation in Unity, Godot, VSeeFace, etc. Multiple model sizes let you balance speed (up to 213 fps) against accuracy. The project includes Unity components for receiving data, visualising points, driving IK, and training a small SVM‑based expression classifier. It works robustly under low light, noise, and partial occlusion, but eye‑region precision is modest and tracking a single face at 30 fps can consume an entire CPU core. Licensed under BSD‑2‑clause.

A real-time 4D view synthesis system that enables high-resolution 4K rendering of dynamic scenes from novel viewpoints.

A JavaScript library for rendering 3D Gaussian Splatting data in the browser, providing a high-level API similar to three.js.

TensorRT‑Alpha is a C++/CUDA toolkit that converts popular CV models (YOLO, EfficientDet, U‑2‑Net, etc.) from PyTorch → ONNX → TensorRT and provides ready‑made inference pipelines with multi‑batch preprocessing, decoding and NMS, targeting Ubuntu 18.04 and Windows 10 GPUs.