10.2312/SRE.20181172
Deep Hybrid Real and Synthetic Training for Intrinsic Decomposition
Bi, Sai
Kalantari, Nima Khademi
Ramamoorthi, Ravi
The Eurographics Association
2018
978-3-03868-068-0
1727-3463
https://diglib.eg.org/bitstream/handle/10.2312/sre.20181172/053-063.pdf
11 pages
Intrinsic image decomposition is the process of separating the reflectance and shading layers of an image, which is a challenging and underdetermined problem. In this paper, we propose to systematically address this problem using a deep convolutional neural network (CNN). Although deep learning (DL) has been recently used to handle this application, the current DL methods train the network only on synthetic images as obtaining ground truth reflectance and shading for real images is difficult. Therefore, these methods fail to produce reasonable results on real images and often perform worse than the non-DL techniques. We overcome this limitation by proposing a novel hybrid approach to train our network on both synthetic and real images. Specifically, in addition to directly supervising the network using synthetic images, we train the network by enforcing it to produce the same reflectance for a pair of images of the same real-world scene with different illuminations. Furthermore, we improve the results by incorporating a bilateral solver layer into our system during both training and test stages. Experimental results show that our approach produces better results than the state-of-the-art DL and non-DL methods on various synthetic and real datasets both visually and numerically.
Eurographics Symposium on Rendering - Experimental Ideas & Implementations
Image-based Techniques
53
63
Sai Bi, Nima Khademi Kalantari, and Ravi Ramamoorthi
CCS Concepts: Computing methodologies → Computer graphics; Computer vision problems; Image processing; Neural networks
Computing methodologies → Computer graphics
Computer vision problems
Image processing
Neural networks
Paper
053-063