Research & Education

3DETR (3D DEtection TRansformer) is an open-source PyTorch implementation and model for 3D object detection, developed by facebookresearch. It presents a simpler alternative to traditional, complex 3D detection pipelines by not relying on 3D backbones like PointNet++ and utilizing minimal 3D-specific operators. The tool achieves comparable or superior performance to existing 3D detection methods such as VoteNet. Its encoder can also be adapted for other 3D tasks, including shape classification. The project provides pretrained model weights and metrics for datasets like SUN RGB-D and ScanNet, along with detailed instructions for installation, dataset preparation, testing, and training.

SmartOne.ai offers specialized real-world AI data services, including annotation, evaluation, and domain expertise for systems that operate outside controlled lab environments. The platform leverages advanced automation combined with expert human insight (Human-in-the-Loop) to precisely categorize, classify, and annotate various data types such as text, images, audio, and video. Key services include Physical AI Data Annotation for embodied intelligence, Data Labeling for AI across industries, and Synthetic Data generation that mirrors real-world conditions without sensitive data challenges. SmartOne.ai also provides Smart Edge Specialists to handle complex technical challenges in AI, IoT, and next-generation technologies, empowering teams to focus on strategy.

U-Mamba is an open-source tool designed to enhance long-range dependency for biomedical image segmentation, built upon the popular nnU-Net framework. It provides a comprehensive repository for researchers and developers focused on deep learning applications within the biomedical field. The tool supports Pytorch 2.0.1 and Mamba, facilitating the implementation of advanced image analysis techniques. Users can train both 2D and 3D models, including U-Mamba_Bot and U-Mamba_Enc, on their own datasets by following provided guidelines for data preparation and preprocessing. It also offers functionalities for inference on testing cases and allows for flexible path settings for existing nnU-Net setups.

Coulomb AI is a data-centric AI platform designed to significantly boost battery performance and extend their operational lifespan. It provides comprehensive real-time analytics on how batteries are performing, identifying key factors that contribute to failure and downtime. The platform suggests corrective actions to avoid sudden breakdowns and ensure 100% uptime. Beyond performance, Coulomb AI focuses on sustainability by uncovering stress factors affecting battery health, preventing untimely aging, and reducing the total cost of ownership. It also facilitates the transition to second-life applications through precise residual value determination. The platform offers a unified analytics suite, integrating all necessary data to monitor, analyze, and optimize batteries across their entire lifecycle, ensuring consistent and high-quality data at any scale.

AICURA medical GmbH offers a medical superintelligence platform designed to scale clinical AI for mission-critical enterprise use cases. The platform aims to unlock significant outcomes, ranging from $50M to $10B+, by leveraging advanced AI capabilities within the healthcare sector. While specific features are not detailed, the focus is on providing a robust AI foundation for complex medical applications, suggesting capabilities in data analysis, predictive modeling, and operational efficiency for large-scale healthcare operations. The company emphasizes its role in powering essential clinical AI, indicating a focus on reliability and high-impact applications.

Peaxy, Inc. offers digital twin solutions specifically designed for critical assets across various industries, including energy, defense, and infrastructure. The platform delivers real-time insights, predictive diagnostics, and lifecycle intelligence, enabling organizations to optimize performance, predict failures, and extend asset lifespan. Key applications include battery analytics and health monitoring, mission readiness for defense systems, and smart facility intelligence for energy insight. Peaxy's approach involves threading data across design, operations, testing, and recycling, integrating live sensor data with advanced analytics models to provide comprehensive diagnostics, performance forecasting, and compliance readiness.

VRIFY is the mining industry's only Exploration Intelligence Software, designed to empower geoscience teams with advanced tools for communication and geological insight. The platform leverages AI-assisted mineral discovery, geospatial data analysis, and 3D visualization to transform complex data into actionable discoveries. Key features include AI prospectivity mapping, 360° imagery, interactive presentations, and data augmentation modules. VRIFY also offers specialized services like DORA for AI-assisted mineral system targeting and Viz for interactive 3D visualization and stakeholder communications. It integrates drillhole, geophysics, and geochemistry data, providing a comprehensive solution for mineral exploration and investor relations.

Workbench Energy offers comprehensive energy management solutions for energy storage systems, power generators, and large power consumers. The platform provides trusted third-party remote operations, real-time monitoring, and market operations for Battery Energy Storage System (BESS) projects, ensuring compliance, responsiveness, and optimization. For large power consumers, it manages demand-side programs like demand response and offers peak prediction services to reduce energy costs. Generators benefit from tailored energy strategies, 24/7 execution, and a market dashboard with real-time electricity system data and private facility reports. Workbench Energy aims to maximize uptime, provide full visibility into assets, and deliver clear, data-driven performance reporting.

hema.to GmbH offers AI-driven cytometry solutions designed to streamline laboratory workflows, enhance reproducibility, and provide greater flexibility. Their flagship product, hema.to CellStudio (RUO), is a browser-based, AI-supported application for flow cytometry analysis that processes anonymized hematology cases, identifies cell populations, and supports automated reporting while maintaining expert control. The platform is cloud-based, offering easy access, automatic updates, and scalability without hardware investment. hema.to also provides a CE-IVD marked solution for B-NHL diagnostic support. The system is privacy-first, GDPR-validated, and ensures data anonymization and encryption, with all data stored within EU boundaries. It boasts high accuracy, with over 4,000 analyses supported and 600,000 FCS files in its database, and has been validated through multiple peer-reviewed publications and clinical studies.

bioloGPT is a specialized scientific data search engine designed to provide researchers and students with results directly from experimental data, rather than relying on abstracts. This platform aims to fuel discoveries by offering a more direct and reliable source of scientific information. Key features include the ability to search raw scientific data, upload files for analysis, and review papers with visual evidence maps. It covers various scientific domains such as Neuroscience, RNA Biology, Immunity, Cancer Microenvironment, Cell Biology, and Evolution, with featured raw findings updated regularly. Users can also sign up for custom summaries of cutting-edge science research.

The Power Electronics Group at KU Leuven – EnergyVille is dedicated to advancing power electronics for a sustainable future. Their research encompasses areas such as magnetic component design, the application of wide bandgap semiconductors, and the development of AI-based design automation tools like PowerBrain. Beyond core power electronics, the group also focuses on critical applications including renewable energy systems, electric mobility, and leveraging AI for forecasting in energy policy. This interdisciplinary approach aims to address contemporary energy challenges through innovative research and educational initiatives, contributing to both academic knowledge and practical solutions.

Beaker is an autonomous AI science journalist that independently builds and writes its own blog, beaker.blog. It focuses on science journalism with genuine excitement, delivering real discoveries and honest reporting. The AI generates articles on a wide range of scientific topics including genetics, biology, physics, paleontology, neuroscience, and more, often citing recent studies and breakthroughs. The platform aims to make complex scientific concepts accessible and engaging for a broad audience, providing detailed yet understandable explanations of new research and its implications.

Quantistry offers a Materials Intelligence Platform designed to accelerate materials discovery and development for R&D teams. The platform leverages AI for materials design and advanced materials simulation software, enabling users to predict material performance before physical synthesis. It integrates quantum mechanics, classical physics, and advanced ML/AI models to determine optimal computational pathways and analyze millions of candidates. This approach helps reduce testing costs and speeds up innovation in areas like battery R&D and alloy design. Quantistry acts as an intelligent co-pilot, automating complex simulations and analyses, allowing scientists and engineers to focus on innovation and achieve faster, data-driven results with a low error rate.

1QBit specializes in solving demanding computational challenges by developing new forms of computational hardware co-developed with software. The company focuses on recasting problems to leverage advanced computing, including quantum computation. They provide automated quantum design tools aimed at utility-scale quantum computation. 1QBit participates in programs like DARPA’s QBI, guiding quantum computing towards utility scale compute. Their approach involves revisiting old problems with next-generation computational tools to discover better solutions and uncover new opportunities.

AuraML is building the core infrastructure for robotics development through its Multimodal World Model (MMWM) for generative simulations. This platform enables faster iteration, safer deployment, and scalable AI integration across industries like logistics, industrial automation, defense, aerospace, agriculture, robotics, and energy. AuraSIM, powered by MMWM, allows robots to perceive, predict, plan, and transfer knowledge. It features a predictive model that generates up to 60 seconds of realistic future simulations from sensor inputs, a multi-control transfer model for adapting tasks across diverse environments, and a customizable reasoning engine for goal-driven cognition. AuraML aims to bridge the sim-to-real gap by training robots in probabilistic worlds, learning to adapt and collaborate before real-world deployment.

The Australian Centre for Robotic Vision is a leading research institution dedicated to advancing the field of robotic vision. It focuses on developing technologies that empower robots to perceive, understand, and interact with their surroundings. Key research areas include sensing, understanding, acting, and learning, with applications spanning environmental monitoring, healthcare, and sustainable food production. The Centre provides resources such as a Robotic Vision Resources Hub, CloudVis for applying vision algorithms, and a wealth of scientific publications and annual reports. It also offers PhD opportunities and fosters industry and academic collaborations, making it a central hub for innovation in robotics in Australia.

Ubotica Technologies specializes in Cognitive Earth Observation, upgrading the entire Earth-Observation stack with intelligence. Their SPACE:AI technology allows for real-time insights and onboard data processing, delivering decision-ready information in minutes. The platform offers Live Maritime Intelligence, a subscription-based predictive surveillance system that detects threats and tracks suspicious vessels. Ubotica's technology is proven with 11 successful SPACE:AI missions and over 30 AI models running in harsh compute environments, validated by organizations like ESA, NASA JPL, and commercial operators. This enables dynamic threat detection and intelligent tasking for Earth observation.

Voltai is at the forefront of developing advanced AI models and agents designed to interact with and understand the physical world. The company's initial focus is on hardware, particularly electronics systems and semiconductors, where AI can push design capabilities beyond human cognitive limits. Voltai's approach involves creating 'world models' that enable agents to run billions of virtual experiments, overcoming the practical limitations of physical testing. These virtual results are then verified in real-world settings to collect rich datasets, mirroring how human experts refine their skills through practical engineering feedback. Voltai collaborates with major semiconductor, consumer electronics, and automotive companies, positioning itself as a strategic partner and a board-level priority for its customers.

AlphaGenome, developed by Google DeepMind, offers a unifying genomics model designed to decipher DNA function. This advanced AI tool is specifically tailored for genomic analysis, assisting researchers and scientists in gaining deeper insights into complex biological systems. While the website content primarily focuses on its core capability of DNA function deciphering, the broader context of DeepMind's work in AI for science suggests its potential applications in areas like protein structure prediction and drug discovery, as mentioned in the previous description. The tool aims to provide a comprehensive approach to understanding genomics, making it a valuable asset for scientific research.

Catched is a comprehensive platform designed for users looking to backorder premium domain names and participate in expired domain auctions. The tool offers extensive registry coverage for both ccTLDs and gTLDs, providing a wide selection of domain options. Users can access detailed TLD information and pricing to make informed decisions. Catched aims to simplify the process of acquiring valuable domain names through its auction and backordering services, catering to individuals and businesses interested in securing specific web addresses.

InSilicoTrials offers an advanced end-to-end in silico platform designed to revolutionize drug development by integrating mechanistic models, AI, digital twins, synthetic data, and operations forecasting. The platform supports both scientific and operational aspects of clinical trials, enabling synthetic patient generation, scenario comparisons for dosing, sample size, and adaptive designs. It includes a Model Library, Data Library, data Integrator, Workflow Builder, Clinical Trial Simulator, and Output Dashboard, supporting various models like safety, efficacy, biomarkers, and AI/ML. A key differentiator is IRIS, a multi-agentic simulation workflow orchestration system that leverages AI to search external sources, interpret data, make recommendations, and independently launch new simulations, significantly de-risking every stage from preclinical to post-approval.

Innophore is an Enzyme & Drug Discovery platform that leverages AI-guided point-cloud technology, specifically its unique Catalophore™ platform, to identify and optimize protein candidates. The tool offers tailor-made solutions for biotechnology and pharmaceutical industries, enhancing standard approaches based on structural or sequence information. It helps formulate problems within its solution space, augmenting R&D pipelines and redefining how potential solutions are approached. Innophore's technology has been applied in large drug-repurposing screenings, such as during the Covid-19 pandemic, demonstrating its capability to find matches beyond conventional search methods.

FedLab is a highly customizable and flexible Federated Learning Framework built on PyTorch, designed to simplify research in this burgeoning area of machine learning. It provides essential modules for FL simulation, including communication, compression, model optimization, and data partition, allowing users to construct FL environments with custom components. FedLab supports various federated learning paradigms and algorithms, offering reproductions of baseline algorithms like FedAvg and FedProx. It also provides extensive data partitioning schemes for both IID and non-IID scenarios, along with visualization tools, making it easier for researchers to manage and experiment with diverse datasets.

Kappazeta offers actionable land monitoring solutions powered by AI-driven satellite insights. It enables governments and businesses to make informed decisions and efficiently analyze vast amounts of global data. The platform provides products for various sectors including agriculture, defense & intelligence, forestry & sustainability, and earth observation. Key capabilities include verifying subsidy compliance, decreasing field inspections, monitoring vegetation health, enhancing situational awareness, detecting deforestation, and improving regulatory compliance and carbon sequestration reporting. Kappazeta leverages scientific remote sensing expertise, global coverage, and near real-time and historical data, with a user-centric and flexible approach.