Research Log

In this introduction, Zeineb presents her medical AI project and explains her long-standing interest in the intersection of medicine and informatics. The focus of her work is on developing systems that can detect brain tumors as early as possible, before they become life-threatening. She discusses why early detection is critical, how improved precision and analysis can provide more options for patients, and how AI-based approaches can support better medical decision-making. This video sets the foundation for her project, outlining the motivation, goals, and potential impact of using AI to improve early tumor detection.

This session focuses on understanding the structure and design choices behind Apple’s approach to 3D reconstruction, and how these tools can be applied or extended in future projects.

In this week’s demo, Zeineb shares an update on her work, walking through recent progress on her model and discussing the latest experiments she’s been running. The session focuses on refining the system and evaluating results as the project continues to move forward.

In this demo, Mugi presents an important update to her project by transitioning from a single-camera setup to a dual-camera solution. By introducing a second viewpoint, the system can capture motion and spatial information more accurately, reducing ambiguity and improving overall reliability.

In this demo, Eman walks through recent updates to his robotics project, focusing on improvements to the camera setup and the robot configuration within the Isaac Lab environment. These changes help create a more stable and accurate system for perception, simulation, and future autonomous development.

In this update Eman showcases a major update to his robotics project. The robot can now move autonomously toward rewards with high accuracy, making decisions based on its environment rather than following predefined paths.

Check out Muhammad Eman Aftab ’s work from our Budapest research lab. Eman is one of our AI researchers, working on inspiring research projects and, most importantly, building in public, with the good, the bad, and the ugly!

In this intro video, Kevin presents Cenarius, a project focused on transforming real-world environments into detailed 3D digital spaces. The long-term vision behind Cenarius is to scan areas such as farmlands, construction sites, and open fields, and convert them into interactive, digitalized environments that can be explored and analyzed.

In this demo, Kevin presents a direct, side-by-side comparison of different applications and AI agents specialized in image analysis. He walks through how each agent approaches visual tasks, highlights their strengths and limitations, and shows where certain tools perform better depending on the use case.

In this demo, Eman walks through two key steps in advancing his vision-based control system. First, he calibrates the camera, upgrading the model from a 2D guess-based approach to a proper 3D calibration setup for more accurate perception. He then sets up the Isaac environment, laying the groundwork for testing, simulation, and future autonomous behavior.

David showcases Venato, an AI-powered regulation monitoring service built to help users stay informed without manual searching. In this demo, he gives a walkthrough of the platform and explains how AI is used to collect, structure, and surface regulatory information in a streamlined way.

In this session, Munkhchimeg Sergelen talks about how avatars are created for SignMate and why the underlying motion data must be extremely clean. ASL relies on precise handshapes and movement paths, so noisy or unrefined tracking can quickly break the accuracy of the sign.

In this demo, Mugi continues her exploration of motion-capture tools by testing Plask and FreeMoCap to understand how well they can track ASL gestures. She compares their strengths, limitations, and the challenges they present when used for sign-language animation.

In this demo, Mugi walks through her early experiments in capturing and generating ASL motion for SignMate. She explores several prototype approaches, including MediaPipe pose tracking, RPM, and ASL Mocap datasets, evaluating how each tool handles the complexity of sign-language gestures.

This session focuses on how SignMate transforms ASL gloss into 3D avatar motion. Mugi walks through the motion-generation system, showing how the model interprets gloss and maps it to realistic signing gestures.

In today’s demo, Munkhchimeg Sergelen explains how American Sign Language (ASL) differs from spoken English and why those differences play a critical role in designing SignMate.

In this week’s Lightbloom Lab demo, Zeineb explains why she upgraded her MRI tumor-detection model from ResNet18 to ResNet50. Her goal: build a stronger system capable of identifying tumors more accurately from MRI scans.

In this introductory video, Eman shares the vision behind his project at the Lightbloom AI Lab in Budapest: building a system that can transform simple hand-drawn sketches into fully playable board games.

In this week’s Lightbloom Lab demo, Kevin showcases a breakthrough in 3D environment reconstruction using Gaussian Splatting — a technique that makes high-quality 3D scans possible without expensive computing or specialized hardware.

This week in the Lightbloom Lab, Eman is showcasing a new vision-driven control system that lets a robot move entirely on its own by reading printed signs. By pointing a camera at simple symbols on paper, the system can interpret the command in real time and adjust its movement without any manual input.

Join us inside the Lightbloom Lab in Budapest as Mugi walks through her latest experiment — attempting to adapt our teammate Kevin into a real-time signing avatar.

Join us inside the Lightbloom Lab in Budapest as Kevin demonstrates how to use our newly assembled computer cluster, the shared infrastructure supporting all ongoing projects.

Join us inside the Lightbloom AI Lab in Budapest as our researcher, Eman, demonstrates the latest updates to our Sketch-to-Game Generator.

Join us inside the Lightbloom AI Lab in Budapest as our researcher, Zeineb, demonstrates the latest critical improvements to our Autoscan AI model for early brain tumor prediction.

Meet Mugi, one of our brilliant AI researchers in the Budapest Lab. She is building SignMate to make communication more accessible through speech to sign avatars.