About Us
The main activity of the Neural Information Processing Group (NIPG) is related to artificial intelligence, with a special focus on the combination of GOFAI (Good Old-Fashioned AI) and deep learning, which we call composite AI, as well as their applications in human-computer interaction. We treat deep learning as a "sensor" and GOFAI as higher-order knowledge that we acquire in school. This is essential for our applications where we combine image processing, video analysis, speech, textual information, and physical processes.
The Department's work was also supported by Robert Bosch Kft., which is partly due to the research results of NIPG. We are members of the National Laboratory for Artificial Intelligence; our industrial R&D&I project is MOBOT, which aims to develop the visual system of a robot intended for warehouse inventory. We participate in several "micro-projects" of the large-scale HumanE-AI EU project, and we will soon join an EIT Digital project focusing on fire detection. Previously, we have successfully completed similar projects, including EU grants and works related to the US Air Force, Honda Europe, and Panasonic's PINTL and OWL programs.
Research Interests
- Human-machine interaction
- Applications in diagnostics, treatments, and training, both behavioral and physical
- Detecting and evaluating human-human interactions
- Information fusion, including image-video, speech, and text
- Temporal processes and prediction
Research Methodology
- Use of computers and GPUs
- In-house developed or licensed software tools
- Development and application of deep learning algorithms
- Composite AI: combining deep learning and GOFAI
Research Staff
- 3 senior researchers
- 1 programmer
- 10 PhD students at different levels
Projects
https://nipg.inf.elte.hu/#grants-page
Selected Publications
- L. Kopácsi, B. Baffy, G. Baranyi, J. Skaf, G. Sörös, S. Szeier, A. Lőrincz, D. Sonntag (2023): Cross-viewpoint semantic mapping: integrating human and robot perspectives for improved 3D semantic reconstruction, Sensors [DOI]
- G. Baranyi, B. C. Dos Santos Melício, Z. Gaál, L. Hajder, A. Simonyi, D. Sindely, J. Skaf, O. Dušek, T. Nekvinda, A. Lőrincz (2022): AI technologies for machine supervision and help in a rehabilitation scenario, Multimodal Technologies and Interaction [DOI]
- Á. Fóthi, L. Soorya, A. Lőrincz (2020): The autism palette: combinations of impairments explain the heterogeneity in ASD, Frontiers in Psychiatry [DOI]
- V. Varga, A. Lőrincz (2020): Reducing human efforts in video segmentation annotation with reinforcement learning, Neurocomputing [DOI]
- M. Véges, V. Varga, A. Lőrincz (2019): 3D human pose estimation with siamese equivariant embedding, Neurocomputing [DOI]
Contact
Website: link