Light-Field Depth Estimation

Light-field depth estimation with recurrent generative models and compact multi-view inputs.

Recurrent conditional adversarial network for light-field depth estimation

Overview

Graduate research on depth estimation from compact subsets of light-field views using a recurrent conditional adversarial network. The project reduced input data by approximately 90% compared with the full 9x9 view stack.

Focus

Light-field imaging · Depth estimation · Recurrent cGAN · Multi-view geometry · Compact model inputs