Abstract: 3D scanners makes usually triangulated meshes. The task is to use a neural network on them. Conversion to volumetric data decreases resolution and takes some time. Conversion to a graph can be solution. Computation of convolution has some specifics. Other aspects of the graph neural networks will be discussded, as augmentation, training, batches, loss functions, recognition of individual nodes and edges.
Who: Tomáš Suk, UTIA
When: 10:00 a.m. Friday, April 24
Where: The session will be held in person at the Institute of Information Theory and Automation (UTIA), in room 45 (café). For directions to the institute, please refer to the following link: https://zoi.utia.cas.cz/index.php/contact
Language: Czech
