Understanding when spatial transformer networks do not support invariance, and what to do about it
| Autor: | Lukas Finnveden, Tony Lindeberg, Ylva Jansson |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
Network complexity Computer science Computer Vision and Pattern Recognition (cs.CV) Computer Science - Computer Vision and Pattern Recognition spatial transformer networks 02 engineering and technology invariant neural networks 010501 environmental sciences 01 natural sciences Convolutional neural network Datorseende och robotik (autonoma system) convolutional neural networks 0202 electrical engineering electronic engineering information engineering Practical implications Computer Vision and Robotics (Autonomous Systems) 0105 earth and related environmental sciences Transformer (machine learning model) business.industry deep learning Pattern recognition Spatial transformation Image alignment 020201 artificial intelligence & image processing Artificial intelligence business |
| Zdroj: | ICPR |
| Popis: | Spatial transformer networks (STNs) were designed to enable convolutional neural networks (CNNs) to learn invariance to image transformations. STNs were originally proposed to transform CNN feature maps as well as input images. This enables the use of more complex features when predicting transformation parameters. However, since STNs perform a purely spatial transformation, they do not, in the general case, have the ability to align the feature maps of a transformed image with those of its original. STNs are therefore unable to support invariance when transforming CNN feature maps. We present a simple proof for this and study the practical implications, showing that this inability is coupled with decreased classification accuracy. We therefore investigate alternative STN architectures that make use of complex features. We find that while deeper localization networks are difficult to train, localization networks that share parameters with the classification network remain stable as they grow deeper, which allows for higher classification accuracy on difficult datasets. Finally, we explore the interaction between localization network complexity and iterative image alignment. Comment: 13 pages, 7 figures, 6 tables |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|