Controllable Data Augmentation Through Deep Relighting
arXiv:2110.13996
Abstract
At the heart of the success of deep learning is the quality of the data. Through data augmentation, one can train models with better generalization capabilities and thus achieve greater results in their field of interest. In this work, we explore how to augment a varied set of image datasets through relighting so as to improve the ability of existing models to be invariant to illumination changes, namely for learned descriptors. We develop a tool, based on an encoder-decoder network, that is able to quickly generate multiple variations of the illumination of various input scenes whilst also allowing the user to define parameters such as the angle of incidence and intensity. We demonstrate that by training models on datasets that have been augmented with our pipeline, it is possible to achieve higher performance on localization benchmarks.
References in corpus (11)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Auto-Encoding Variational Bayes
- GAN-based Synthetic Medical Image Augmentation for increased CNN Performance in Liver Lesion Classification
- Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations
- Smart Augmentation - Learning an Optimal Data Augmentation Strategy
- R2D2: Repeatable and Reliable Detector and Descriptor
- Single Image Portrait Relighting
- Improving Deep Learning using Generic Data Augmentation
- VIDIT: Virtual Image Dataset for Illumination Transfer
- Image Stylization for Robust Features
- Disentangling and Learning Robust Representations with Natural Clustering