Semantic segmentation of wide-angle lens images in urban/indoor environments

M-GEO
ACQUAL
Staff Involved
Additional Remarks

Suggested elective courses: Laser scanning, Scene Understanding with Unmanned Aerial Vehicles / other deep learning courses, Programming courses (python)

No fieldwork included/required

Topic description

Data acquired from indoor environments contains a lot of valuable information about the assets and facilities. Here, the idea is to process and analyze images obtained from a Ladybug 5+ 360 degree spherical video camera. See background information: Lehtola, V. V., Nikoohemat, S., & Nüchter, A. (2021). Indoor 3D: Overview on scanning and reconstruction methods. Handbook of Big Geospatial Data, 55-97.

Topic objectives and methodology

Machine/deep learning methods for classification/segmentation applied on wide-angle lens images. Camera calibration tasks are included. Programming is done preferably with python3.

References for further reading

Lehtola, V. V., Nikoohemat, S., & Nüchter, A. (2021). Indoor 3D: Overview on scanning and reconstruction methods. Handbook of Big Geospatial Data, 55-97.