Research map / 3D vision / Depth and stereo
FourierNets enable the design of highly non-local optical encoders for computational imaging
Diptodip Deb, Zhenfei Jiao, Ruth R Sims, Alex Bo-Yuan Chen, Michael Broxton, Misha Ahrens, Kaspar Podgorski, Srinivas C Turaga
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 |
|---|---|---|---|
| Use global Fourier convolutions rather than the UNet decoder | DiffuserCam lensless photography reconstruction · DiffuserCam Lensless Mirflickr Dataset test | MSE (×10^-2): 1.68 ± 0.060 -> 0.39 ± 0.007 | Improved |
| Use global Fourier convolutions rather than Le-ADMM-U | DiffuserCam lensless photography reconstruction · DiffuserCam Lensless Mirflickr Dataset test | MSE (×10^-2): 0.75 ± 0.021 -> 0.39 ± 0.007 | Improved |
| Train FourierNet with MSE+LPIPS instead of MSE loss (ablation) | FourierNet · DiffuserCam Lensless Mirflickr Dataset test | LPIPS: 0.20 ± 0.00 -> 0.16 ± 0.00 | Improved |
| Train FourierNet with MSE+LPIPS instead of MSE loss (ablation) | FourierNet · DiffuserCam Lensless Mirflickr Dataset test | MSE (×10^-2): 0.39 ± 0.007 -> 0.54 ± 0.010 | Hurt |
| Use FourierNet2D rather than UNet2D to optimize the microscope encoding | Zebrafish snapshot microscopy · Zebrafish snapshot microscopy, 256×256 camera | L_HNMSE: 0.7109 ± 0.0161 -> 0.6093 ± 0.0209 | Improved |
| Use FourierNet3D rather than UNet3D for reconstruction | FourierNet2D-optimized microscope · Zebrafish snapshot microscopy, 256×256 camera | L_HNMSE: 0.7298 ± 0.0151 -> 0.6093 ± 0.0209 | Improved |
| Add Wiener deconvolution before the UNet | UNet-based microscopy decoder · Zebrafish snapshot microscopy, 256×256 camera | L_HNMSE: 0.7109 ± 0.0161 -> 0.7223 ± 0.0179 | Mixed |
| Add multiscale Fourier features to FourierNet (ablation) | FourierNet lensless reconstruction · DiffuserCam Lensless Mirflickr Dataset test | MSE (×10^-2): 0.39 ± 0.007 -> 0.43 ± 0.009 | Hurt |
| Add multiscale Fourier features to FourierNet (ablation) | FourierNet 3D volume reconstruction · Zebrafish snapshot microscopy, 256×256 camera | PSNR: 34.89 ± 0.88 -> 34.87 ± 0.82 | No clear effect |
| Optimize microscope parameters for the same ROI type used at test time | Programmable zebrafish snapshot microscope · Type A ROI | PSNR: 46.03 ± 1.33 -> 49.76 ± 1.35 | Improved |
| Optimize microscope parameters for the same ROI type used at test time | Programmable zebrafish snapshot microscope · Type B ROI | PSNR: 35.56 ± 1.41 -> 37.34 ± 0.96 | Improved |
| Optimize microscope parameters for the same ROI type used at test time | Programmable zebrafish snapshot microscope · Type C ROI | PSNR: 30.87 ± 1.15 -> 33.79 ± 0.90 | Improved |
Closest papers by meaning
- Fast Fourier Convolution (NeurIPS 2020 · similarity 0.86)
- Photon-Efficient 3D Imaging with A Non-Local Neural Network (ECCV 2020 · similarity 0.86)
- Improving the Learning Capability of Small-size Image Restoration Network by Deep Fourier Shifting (NeurIPS 2024 · similarity 0.85)
- All-Optical Nonlinear Diffractive Deep Network for Ultrafast Image Denoising (CVPR 2025 · similarity 0.84)
- Neural Tangent Knowledge Distillation for Optical Convolutional Networks (NeurIPS 2025 · similarity 0.83)
- Learning Spatially Collaged Fourier Bases for Implicit Neural Representation (AAAI 2024 · similarity 0.83)
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.
