Research map / 3D vision / Depth and stereo
Neural EPI-Volume Networks for Shape From Light Field
Stefan Heber, Wei Yu, Thomas Pock
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| extend input from single 2D EPIs to EPI volumes with 3D convolutions | Heber [14] u-shaped network · Synthetic POV-Ray LF dataset (50 test LFs) | RMSE: 0.80 -> 0.83 | No clear effect |
| extend input from 2D EPIs to EPI volumes with 3D convolutions | Heber [14] u-shaped network · Synthetic POV-Ray LF dataset (50 test LFs) | MAE: 0.35 -> 0.34 | Improved |
| extend input from 2D EPIs to EPI volumes with 3D convolutions | Heber [14] u-shaped network · Synthetic POV-Ray LF dataset (50 test LFs) | pixels with relative error > 0.2% (%): 14.76 -> 14.42 | Improved |
| extend input from 2D EPIs to EPI volumes with 3D convolutions | Heber [14] u-shaped network · Synthetic POV-Ray LF dataset (50 test LFs) | computation time: 1.3s -> 0.8s | Improved |
| extend input from 2D EPIs to EPI volumes with 3D convolutions | Heber [14] u-shaped network · Synthetic and SLFA light fields | streaking artifacts | Improved |
| use CNN on whole EPI volumes instead of patch-based CNN (Heber [12], no refinement) | patch-based CNN · Synthetic POV-Ray LF dataset (50 test LFs) | RMSE: 1.87 -> 0.83 | Improved |
| use learned network instead of non-learning Jeon [17] | Jeon [17] · Synthetic POV-Ray LF dataset (50 test LFs) | computation time: 2h 12min 30s -> 0.8s | Improved |
Closest papers by meaning
- EPINET: A Fully-Convolutional Neural Network Using Epipolar Geometry for Depth From Light Field Images (CVPR 2018 · similarity 0.91)
- Light Field Reconstruction Using Deep Convolutional Network on EPI (CVPR 2017 · similarity 0.91)
- Light Field Reconstruction Using Convolutional Network on EPI and Extended Applications (TPAMI 2018 · similarity 0.91)
- Light Field Intrinsics With a Deep Encoder-Decoder Network (CVPR 2018 · similarity 0.90)
- Convolutional Networks for Shape From Light Field (CVPR 2016 · similarity 0.89)
- Light Field Networks: Neural Scene Representations with Single-Evaluation Rendering (NeurIPS 2021 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
