Thesis supervisor: Bence Molnár
Location of studies (in Hungarian): BME Fotogrammetria és Térinformatika Tanszék Abbreviation of location of studies: EOFT
Description of the research topic:
Processing a point cloud faces significant challenges for engineers, whether it was collected by photogrammetry or laser scanning. The use of point clouds in engineering is widespread because it contains a large amount of information about the object to be measured, providing geometric data and enabling visualization through color information. At the same time, the large amount of data also makes storage and processing complex. Currently, the interpretation of data can mostly only be solved with the help of human intervention.
The possibility of applying artificial intelligence has opened up new perspectives in spatial data processing based on photographs. Similar progress is expected in the processing of point clouds with the support of AI.
The research aims to support point cloud processing with deep learning. With the creation of a neural network supporting the processing of point clouds, the thinning of the point quantity can be achieved in the first step; in the further steps, classification and object recognition can be the goal.
Preliminary research plan:
1. A thorough review of state-of-the-art literature and applications of deep learning algorithms applied on images and spatial data.
2. Select the nature and area of use of the objects to be examined during the research (e.g., structural inspection, industrial automation, autonomous driving vehicles, medical diagnostics); collect data that you can use as training and validation data in the research.
3. Develop one or more efficient neural networks that represent advances in data processing compared to current market solutions.
4. Carry out a qualification process that allows to compare and evaluate the own results against other solutions on the field.
During the research, potential cooperation is available with other BME Faculties, other departments of the Faculty of Civil Engineering, and with the industrial partners of the Department of Photogrammetry and Geoinformatics.
Deadline for application: 2022-12-20
2024. IV. 17. ODT ülés Az ODT következő ülésére 2024. június 14-én, pénteken 10.00 órakor kerül sor a Semmelweis Egyetem Szenátusi termében (Bp. Üllői út 26. I. emelet).
Login
