Learning to Estimate Indoor Lighting from 3D Objects

  • 2018-06-12 17:06:19
  • Henrique Weber, Donald Prévost, Jean-François Lalonde
  • 1

Abstract

In this work, we propose a step towards a more accurate prediction of theenvironment light given a single picture of a known object. To achieve this, wedeveloped a deep learning method that is able to encode the latent space ofindoor lighting using few parameters and that is trained on a database ofenvironment maps. This latent space is then used to generate predictions of thelight that are both more realistic and accurate than previous methods. Toachieve this, our first contribution is a deep autoencoder which is capable oflearning the feature space that compactly models lighting. Our secondcontribution is a convolutional neural network that predicts the light from asingle image of a known object. To train these networks, our third contributionis a novel dataset that contains 21,000 HDR indoor environment maps. Theresults indicate that the predictor can generate plausible lighting estimationseven from diffuse objects.

 

Introduction (beta)

None

 

Conclusion (beta)

None