Deep Coded Aperture Design: An End-to-End Approach for Computational Imaging Tasks

This repository provides the Python source codes related to the paper "Deep Coded Aperture Design: An End-to-EndApproach for Computational Imaging Tasks"

Installation

List of libraries required to execute the code.:

• python = 3.7.7
• Tensorflow = 2.2
• Keras = 2.4.3
• numpy
• scipy
• matplotlib
• h5py = 2.10
• opencv = 4.10
• poppy = 0.91

All of them can be installed via conda (anaconda), e.g.

conda install jupyter

or using pip install and the required file.

Data

This work uses the following three datasets. Please download the datasets and store them it correctly in the corresponding dataset folder (Train/Test).

• MNIST dataset: Provided in the dataset/MNIST folder.
• ARAD hyperspectral dataset: It contains 450 hyperspectral training images and 10 validation images. The dataset is available on the challenge track websites. Note that registration is required to access data.
• NYU Depth Dataset: It contains 1449 RGB images. We use a depth map of 15 discretization levels and its semantic labels for 13 classes. A Matlab function to convert to 15 discretization levels is provided in the dataset/NYU folder.

Structure of directories

Directory	Description
Dataset	Folder that contains the datasets
Models and Tools	.py files for the custumer models