Simpler is better: spectral regularization and up-sampling techniques for variational autoencoders

no code implementations • 19 Jan 2022 • Sara Björk, Jonas Nordhaug Myhre, Thomas Haugland Johansen

The current solution to avoid this is to either replace transposed convolutions with bilinear up-sampling or add a spectral regularization term in the generator.

Instance Segmentation of Microscopic Foraminifera

1 code implementation • 15 May 2021 • Thomas Haugland Johansen, Steffen Aagaard Sørensen, Kajsa Møllersen, Fred Godtliebsen

The model achieves a (COCO-style) average precision of $0. 78 \pm 0. 00$ on the classification and detection task, and $0. 80 \pm 0. 00$ on the segmentation task.

Instance Segmentation Novel Object Detection +4

Towards detection and classification of microscopic foraminifera using transfer learning

1 code implementation • 14 Jan 2020 • Thomas Haugland Johansen, Steffen Aagaard Sørensen

Foraminifera are single-celled marine organisms, which may have a planktic or benthic lifestyle.

Classification Image Classification +1