Deep Hybrid Real and Synthetic Training for Intrinsic Decomposition

TL;DR

This paper introduces a hybrid deep learning approach for intrinsic image decomposition that trains on both synthetic and real images, utilizing a novel supervision method and a bilateral solver to improve results on real-world data.

Contribution

It presents a new hybrid training strategy combining synthetic and real images, along with a bilateral solver layer, to enhance intrinsic decomposition performance.

Findings

Outperforms state-of-the-art methods on synthetic datasets

Achieves better results on real images both visually and numerically

Demonstrates the effectiveness of hybrid training and bilateral solver layers

Abstract

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.