A Python library for ensembling bounding box predictions from multiple object detection models using methods like Weighted Boxes Fusion (WBF) and NMS.

torchdistill is a PyTorch‑based library that lets you run knowledge‑distillation experiments (teacher‑student training) by writing only a YAML configuration. It provides many modern KD losses, a forward‑hook manager for extracting intermediate features, and example scripts for vision and NLP tasks. Install via `pip install torchdistill` and cite the associated papers when used in research.

A PyTorch implementation of AlphaFold 3 for predicting the 3D structures of biomolecular interactions, including comprehensive tools for PDB dataset curation.

A PyTorch Lightning-powered library of self-supervised methods for unsupervised visual representation learning, providing a unified environment to train and compare SOTA visual models.

A TensorFlow-based project that trains an AI agent to drive Mario Kart 64 by learning from human gameplay recordings.

A collection of Python scripts for converting object detection dataset annotations between CSV, COCO, Pascal VOC, and Labelme formats.

An offline OCR library for Android that enables high-speed recognition of Chinese ID cards, driver's licenses, and passports without requiring an internet connection.

A packaged version of the Segment Anything Model (SAM) that provides easy pip installation and extended support for video segmentation and cloud GPU integration.

A library for point cloud registration using probabilistic models, offering a more robust alternative to ICP for both rigid and non-rigid transformations.

A collection of hardware-aware efficient ConvNet models and tools, including FBNet and ChamNet, designed to optimize computer vision performance and latency on mobile devices.

A comprehensive benchmark and framework for stereo matching that integrates 17 datasets and multiple state-of-the-art models for depth estimation.

A package of trained U-net models for automated lung segmentation in CT scans, capable of handling severe pathologies and lobe-specific masking.

An NPM package that brings the OpenCV computer vision library to Node.js and web browsers, enabling tasks like real-time face detection in JavaScript.

A hierarchical perception library for autonomous systems that provides modular APIs for building computer vision pipelines using TensorFlow and Keras.

A fast, deep-learning based neuroimaging pipeline for volumetric and surface-based thickness analysis of brain MRI images, serving as a high-speed alternative to FreeSurfer.

A high-performance C++ and TensorRT framework that fuses YOLO object detection with monocular/binocular depth estimation to track objects and estimate their motion states.

A whole-slide foundation model for digital pathology that uses a tile-and-slide encoder architecture to analyze massive pathology images.

A comprehensive TensorFlow Keras pipeline for YOLOv2, v3, and v4 object detection, supporting diverse backbones, advanced training techniques, and on-device deployment.

A deep-learning framework for multi-animal pose tracking that uses a human-in-the-loop GUI to rapidly label and quantify animal behavior.

A high-performance tool for converting 3D Gaussian Splatting files between multiple formats with GPU-accelerated filtering and compression.

A facial recognition app for Nextcloud that detects, analyzes, and groups faces in images locally to provide private, person-based photo organization.

A React Native demo and integration guide for an offline face recognition SDK that supports 1:1 verification and liveness detection on iOS and Android.

A Python library that allows users to call the local OCR model built into the WeChat Windows client to perform text recognition on images.

An implementation of Vision Mamba, a bidirectional state space model for efficient visual representation learning that is significantly faster and more memory-efficient than traditional Vision Transformers.

Pixio is a vision encoder optimized for dense prediction tasks like depth estimation and semantic segmentation, achieved through enhanced pixel reconstruction pre-training.

An Android SDK for secure identification document scanning and data extraction using machine learning and native C++ libraries.

A binocular vision system for camera calibration, disparity mapping, and 3D reconstruction using algorithms like ELAS and ADCensus.

A set of Expo modules that wrap Google's MLKit native libraries, enabling on-device machine learning features like face and object detection in Expo apps.

PE3R is a perception-efficient 3D reconstruction system that converts 2D images into semantically understood 3D scenes using zero-shot generalization.

A multi-task deep learning framework for real-time 2D and 3D human pose estimation and action recognition.

A Streamlit-based vision studio for real-time object detection, segmentation, pose estimation, and tracking using YOLO26 and YOLO World v2.

A multimodal AI project that generates virtual spatial proteomics (mIF) profiles from routine H&E pathology slides for tumor microenvironment modeling.

A deep neural network-based method for automatically picking seismic arrival times (P and S waves) from raw seismic data.

A neural network based algorithm for heterogeneous cryo-EM and cryo-ET reconstruction that models continuous distributions of 3D structures.

A lightweight Python library for analyzing the performance of infrared small target detection models using pixel-level, target-level, and hybrid metrics.

A unified library for Multiple Instance Learning (MIL) in computational pathology, providing a standardized framework to implement and compare numerous MIL architectures for whole-slide image analysis.

An on-device, fully offline SDK for iOS that provides face detection, recognition, and liveness detection to enable secure identity verification without network access.

An iOS SDK for real-time pose estimation and skeleton tracking, providing pre-built tools for exercise counting and range-of-motion analysis.

A ComfyUI wrapper for Meta's SAM 3D Body that enables full-body 3D human mesh recovery from a single image.

An end-to-end animal pose estimation package that uses transformer architectures to robustly track movements in single-view and multi-view videos, especially during occlusions.

A Python framework for whole slide image analysis that integrates with the scverse ecosystem to enable interoperable and scalable histological workflows.

A deep learning library for in-the-wild 3D face reconstruction and animation, featuring tools for speech-driven avatars and emotion recognition.

A PyTorch toolkit for accelerated MRI reconstruction that provides an end-to-end pipeline for sampling, reconstruction, and validation using deep learning.

Mundi is an open source web GIS that uses LLMs to automate geoprocessing algorithms and symbology editing for vector, raster, and point cloud data.

A CNN-based architecture for panoptic segmentation that predicts class- and instance-specific labels for every pixel in an image.

A real-time radiance field rendering project that replaces Gaussian kernels with deformable Beta Kernels to improve geometric detail, color representation, and memory efficiency.

A computer vision project that integrates YOLOv7 and YOLOv8 for real-time object detection and tracking across video streams from various sources.

A large-scale multi-view RGB-D dataset of 10,000 indoor scenes designed for training AI models in 3D reconstruction and semantic segmentation.