Research & Education

PoseLib is an open-source library specifically designed for solving calibrated camera pose estimation problems. It offers minimal solvers that focus on absolute pose estimation from various types of correspondences. The library's primary goal is to provide implementations that are both fast and robust, making it suitable for integration into computer vision and robotics projects. Its open-source nature allows for community contributions and widespread use in academic and industrial settings.

PettingZoo is a Python library specifically designed to support research in multi-agent reinforcement learning. It offers a standardized API, making it easier for researchers and developers to create and evaluate multi-agent environments consistently. The library comes equipped with reference environments and various utilities, streamlining the process of research and development in this complex field. It functions as a multi-agent counterpart to Gymnasium, providing a familiar structure for those accustomed to single-agent reinforcement learning frameworks.

CenterTrack is an open-source computer vision tool that focuses on simultaneous object detection and tracking. Its core methodology involves tracking objects by identifying and following their center points. This tool is primarily designed for use in research and development contexts within the computer vision field, offering a method for efficient object tracking. It is available on GitHub, indicating its open-source nature and accessibility for developers and researchers.

CenterFusion is an open-source tool that provides a center-based implementation for fusing radar and camera data to achieve 3D object detection. It is primarily designed for research and development purposes in fields such as autonomous vehicles and robotics. By combining information from both radar and camera sensors, CenterFusion aims to improve the accuracy and robustness of object detection systems, which is crucial for advanced applications in these domains. The project is available on GitHub, indicating its open-source nature and accessibility for developers and researchers.

boxx is an open-source Python toolbox specifically designed to enhance efficiency in building and debugging applications, particularly within the domains of scientific computing and computer vision. It provides a comprehensive collection of utilities and functions aimed at simplifying complex tasks and optimizing development workflows. The tool's primary goal is to boost the productivity of Python developers by offering readily available solutions for common challenges in these specialized fields.

VAD is an open-source tool specifically designed for vectorized scene representation, aiming to enhance efficiency in autonomous driving systems. It provides a robust method for modeling complex environments by utilizing vectorized data, which can lead to more streamlined and accurate navigation. This tool is primarily intended to support research and development efforts in the field of AI-driven navigation and autonomous vehicle technology. Its availability on GitHub facilitates collaboration and further innovation within the community.

Materials Explorer is an AI-powered research tool specifically developed for the field of materials science. Hosted on Hugging Face, it provides capabilities for exploring materials data and conducting scientific analysis. The tool is designed to support both advanced research endeavors and educational applications, making complex materials science data more accessible and interpretable for a diverse audience.

PoseDiffusion is an open-source project focused on advancing pose estimation techniques. It utilizes a novel approach called diffusion-aided bundle adjustment to improve the accuracy and robustness of pose estimation. The tool provides access to the research, code, and related resources, making it valuable for those working in the field of computer vision. It is specifically designed for AI researchers and computer vision engineers who are interested in exploring and implementing cutting-edge pose estimation methodologies.

Artificial, Inc. provides the Artificial Product Suite, a cloud-based software platform specifically designed for life science laboratories. This platform facilitates the connection and orchestration of various lab instruments, devices, and databases, streamlining lab operations. Key functionalities include real-time process control and monitoring, ensuring efficient and accurate experimental execution. The suite is also adept at capturing AI-ready data, which can be leveraged for advanced analytics and insights, while maintaining human oversight and intervention where necessary.

The TensorFlow Object Detection API is a robust framework designed for the creation, training, and deployment of object detection models. As an integral part of the broader TensorFlow ecosystem, it provides developers with the necessary tools to build sophisticated AI models capable of identifying and precisely locating various objects within visual data, including both still images and video streams. This API simplifies the complex process of developing computer vision applications focused on object recognition.

DifferentialEquations.jl is a comprehensive, multi-language software suite engineered for high-performance numerical solutions of differential equations. It provides robust solvers for a wide array of equation types, including ordinary differential equations (ODEs), stochastic differential equations (SDEs), delay differential equations (DDEs), and differential-algebraic equations (DAEs). The tool is specifically designed to integrate with scientific machine learning (SciML) components, offering powerful capabilities for researchers and developers in computational science. It is primarily implemented in the Julia programming language, leveraging its performance advantages.

Enzyme QMS is a specialized quality management system (QMS) software tailored for the life sciences industry. It is designed to help companies adhere to critical regulatory standards, including cGMP, QSR, and ISO. The software provides comprehensive support across all phases of the product development lifecycle, ensuring that quality and compliance requirements are consistently met. This tool aims to streamline quality processes and documentation for life sciences organizations.

Malib is designed as a parallel framework specifically for population-based multi-agent reinforcement learning (MARL). Its core purpose is to support the development and implementation of complex multi-agent systems. The tool provides the necessary infrastructure to apply reinforcement learning techniques within a population-based learning paradigm, enabling researchers and developers to explore and optimize multi-agent behaviors and interactions.

contrastors is a specialized toolkit designed for the development and assessment of contrastive learning models. It leverages Flash Attention to ensure rapid and efficient training processes. The toolkit is engineered to support training across multiple GPUs, enhancing its performance capabilities. Additionally, contrastors incorporates GradCache support, which is crucial for handling large batch sizes effectively, even in environments with limited memory resources. This makes it suitable for researchers and developers working on advanced machine learning tasks.

MCUNet is a collection of compact deep learning models specifically engineered for Internet of Things (IoT) devices. Its core focus is on achieving highly memory-efficient, patch-based inference, allowing complex AI tasks to run directly on hardware with limited computational and memory resources. Furthermore, MCUNet supports on-device training even under stringent memory constraints, making it suitable for applications requiring adaptive learning without constant cloud connectivity. This technology empowers developers to deploy sophisticated AI capabilities to edge devices that would otherwise be unable to handle such workloads.

diffeqpy is a Python package that provides robust capabilities for solving various types of differential equations. It integrates with DifferentialEquations.jl, a high-performance Julia library, to power its core routines. This integration allows diffeqpy to offer efficient and accurate solutions for complex mathematical problems. The tool is particularly useful for applications in scientific machine learning, where differential equations are fundamental, and for general mathematical modeling tasks across different scientific and engineering disciplines.

deeponet is a specialized tool designed for learning nonlinear operators, leveraging the DeepONet architecture. It offers the source code associated with a research paper focused on the universal approximation theorem of operators. This tool is particularly relevant for researchers and practitioners in scientific computing who need to model complex nonlinear relationships. Its core utility lies in providing a foundational implementation for advanced operator learning tasks.

Optimization Engine (OpEn) is a powerful solver specifically engineered for embedded optimization tasks, delivering both speed and accuracy. Its primary application lies in next-generation robotics and autonomous systems, where efficient and reliable optimization is critical. The tool facilitates code generation, enabling real-time optimization capabilities, and offers robust integration with ROS (Robot Operating System). OpEn is particularly well-suited for tackling nonconvex optimization problems, providing a versatile solution for complex control and decision-making scenarios in advanced robotic applications.

Open3D-ML is an extension of the Open3D core library, specifically designed to facilitate 3D machine learning tasks. It provides a comprehensive set of tools for processing and analyzing 3D data within a machine learning context. The platform is particularly geared towards applications such as semantic point cloud segmentation, which involves classifying individual points in a 3D point cloud, and object detection in 3D environments. By building upon the robust Open3D framework, Open3D-ML aims to offer a powerful and flexible solution for developers and researchers working with 3D machine learning.

polyrnn-pp-pytorch is a PyTorch-based re-implementation of the Polygon-RNN++ model. This tool provides the functionality to train new Polygon-RNN++ models, allowing researchers and developers to leverage its capabilities for various tasks. It also supports running demonstrations directly on local machines, which is beneficial for testing and development. The primary focus of polyrnn-pp-pytorch is to facilitate efficient and interactive annotation of segmentation datasets, streamlining the process of preparing data for machine learning models.

Nuclear Fusion sim FUSION CIRCUS beta is a browser-based simulation tool designed for exploring nuclear fusion concepts. It accurately models tokamaks such as ITER and JET, which aim to harness sun-like power at extreme temperatures within magnetic fields. Built with machinery technology, the simulator integrates real-world physics, including Bosch-Hale and IPB98 models. It features 28 distinct modules, 16 different tokamaks, and an AI coach to guide users through the simulations.

photon-ml is a machine learning library specifically designed to operate on Apache Spark. This library provides robust support for training Generalized Linear Models (GLMs) and Generalized Linear Mixed Models (GLMMs). Originally developed by LinkedIn, photon-ml is engineered to handle scalable machine learning tasks efficiently. It is particularly well-suited for large-scale data analysis and the training of complex models, making it a valuable tool for big data environments.

best-of-atomistic-machine-learning provides a regularly updated, ranked compilation of open-source projects in atomistic machine learning (AML). This resource is designed to assist researchers and developers in identifying and leveraging various libraries, tools, and other resources pertinent to AML. It serves as a central hub for discovering valuable assets for both the development and research aspects of atomistic machine learning.

ByteNet is a machine translation tool specifically designed for French-to-English translation. It is built using TensorFlow and implements DeepMind's innovative ByteNet architecture. A key feature of ByteNet is its use of dilated and causal conv1d layers, which serve as a replacement for traditional Recurrent Neural Networks (RNNs). This architectural choice contributes to its ability to achieve fast training times and deliver state-of-the-art performance, particularly in character-level translation tasks.