Round Corner
Department of Computer and Information Science


Computer Vision, Deep Learning & Industry (EFI, Trollhetta)

Several projects proposed by industry related to computer vision and deep learning are available (more information can be found at the end):

  • Thermo Fisher Scientific (EFI): Materials & Structural Analysis, e.g. segmentation of CT scan of drilled cylinders
  • Trollhetta: 1) General Deep Learning, 2) Segmentation with emphasis on Deep learning and 3) Safety in nursing homes. (more info can be found below)

 The projects consist of:

  • Studying state-of-the-art solutions within the field
  • Development of prototype software
  • Testing and validation.

Contact the project provider for more details about the project. The focus of the project will be defined together with the student. It's also possible to work in groups of two.


Thermo Fisher Scientific (EFI): 

Digital Rocks technology is based on the use of high resolution 3D images to derive digital models of reservoir rock. These models in combination with the right simulation tools can be used to derive macroscopic rock properties. Imaging processing of these images, filtering and segmentation, is a critical step for obtaining an accurate characterization. The result of these processes is an image volume containing voxels that are individually tagged according to their phase identity (grain, micro-phase, pore). The main goal of this project is to investigate the applicability of Machine Learning (ML) and Deep Learning (DL) techniques in order to develop a coherent workflow to perform automatic segmentation of reservoir rocks. Several 3D X-Ray tomography images and their corresponding segmented images, generated using traditional techniques, will be used to train and test these new methods. 



1) General Deep Learning
Deep learning has during the last couple of years become a favorite method for many applications. The goal of this thesis is to design and implement a flexible and user friendly Deep learning subsystem that can be imbedded into existing systems to augment the areas of applications. The module(s) will be written in C++ and the workbench will be TrollBrain, a general AI framework developed by Trollhetta AS (Contact: Ketil Bø,, Mob.: 90112451)

2) Segmentation with emphasis on Deep learning.
Segmentation of images is a crucial step in the image analysis pipeline that, up to now, has not been solved satisfactory for more challenging scenes. One of the potential solutions may be Deep learning that has some promising qualities in this respect. The goal of the thesis is to investigate new methods for robust segmentation including Deep learning. The subsystem will be written in C++ and the workbench will be TrollView, an image processing framework developed by Trollhetta AS. (Contact: Ketil Bø,, Mob.: 90112451)

3) Safety in nursing homes.
In the future, several tasks in nursing homes will be partly automated both because of demand for increased safety for the inhabitant and to save resources.
The goal of this thesis is to use image analysis and artificial intelligence to, in a discrete way, “keep an eye” on the client to detect unnormal situations while ignoring everything else. The subsystem will be written in C++ and the workbench will be TrollEye, an artificial vision framework developed by Trollhetta AS (Contact: Ketil Bø,, Mob.: 90112451)


Frank Lindseth Frank Lindseth
414 IT-bygget
735 95493 / 928 09 372
NTNU logo