4D Light Field Benchmark

Check out the new Google group to connect with other light field researchers, discuss ideas, and initiate collaborations. Some questions about the benchmark are answered over there as well! There is also a new collection of light field resources. Please feel encouraged to add further links.

go back to table

Acronym

EPI-Refocus-Net

Method title

EPI and Refocus learning for light field depth estimation

Method description

We combine EPI and Refocus clues for light field depth estimation using convolutional neural networks. Refer to our previous work:
" Zhou, W., Liang, L., Zhang, H., Andrew, L., & Lin, L. (2018). Scale and Orientation Aware EPI-Patch Learning for Light Field Depth Estimation. International Conference on Pattern Recognition."
"Luo, Y., Zhou, W., Fang, J., Liang, L., Zhang, H., & Dai, G. (2017). Epi-patch based convolutional neural network for depth estimation on 4d light field. International Conference on Neural Information Processing."

Camera setup

full grid

Parameter description

Using Convolutional Layers and Fully Connected Layers

Programming language

PyTorch + Python

Runtime environment

Intel Core i7-4720HQ 2.60GHz + Two TITAN X GPUs

Additional data

Raw evaluation scores for all scenes