Data Augmentation via Mixing Images


What to expect
This presentation covers recent advances in data augmentation via mixing images for computer vision. You’ll gain a deeper understanding of the two main classes of augmentation: pixel-wise mixing and patch-wise mixing. You’ll also discover what types of problems they are most useful for and the other modalities (i.e., text) they can be applied to. For each augmentation method, you’ll see what the resulting augmented images would look like and learn how to empirically evaluate it.
Data Augmentation for Computer Vision 101
Pixel-Wise Mixing
Patch-Wise Mixing
Empirical Evaluation of Methods
Review past data augmentation methods and become familiar with recent advances.
Find out how pixel-wise mixing is done and where it is most useful.
Learn how patch-wise mixing is done and see the types of problems it solves.
Learn how to assess how much training sets have improved thanks to advancements in data augmentation.
Data Augmentation for Computer Vision 101
Pixel-Wise Mixing
Review past data augmentation methods and become familiar with recent advances.
Find out how pixel-wise mixing is done and where it is most useful.
Patch-Wise Mixing
Empirical Evaluation of Methods
Learn how patch-wise mixing is done and see the types of problems it solves.
Learn how to assess how much training sets have improved thanks to advancements in data augmentation.
Data Augmentation for Computer Vision 101
Review past data augmentation methods and become familiar with recent advances.
Pixel-Wise Mixing
Find out how pixel-wise mixing is done and where it is most useful.
Patch-Wise Mixing
Learn how patch-wise mixing is done and see the types of problems it solves.
Empirical Evaluation of Methods
Learn how to assess how much training sets have improved thanks to advancements in data augmentation.


About Our Speaker
Dominik Lewy has over eight years of hands-on experience in machine learning, deep learning, data exploration, and business analysis projects, primarily in the FMCG industry. He is a technical leader, setting goals and preparing road maps for projects. He is also a PhD candidate at the Warsaw University of Technology where he focuses on the study of neural networks for image processing. He tries to be a bridge between commercial and academic worlds. His main research interest is digital image processing in the context of facilitating adoption of deep learning algorithms in business, where training data is scarce or nonexistent.