Motivation

Compressed sensing is a technique used in signal processing and data compression. It is particularly effective when dealing with signals or data that can be represented sparsely. It can provide near-perfect reconstruction with randomly sampled data.

In a task-aware setup, the goal is to replace random sampling with optimization-based sampling. In addition, deep learning-based methods will be utilized as they have shown superior performance in data reconstruction in comparison to traditional reconstruction methods.

Objectives

Implementing a functioning deep learning-based compressed sensing framework

Statistical evaluation and comparison of the proposed method with classical CS methods

Requirements

Familiar with artificial neural networks and Python programming language

Interest in image and signal processing

Basic knowledge of compressed sensing.

Literature

[1] .B. Chen and J. Zhang, "Content-aware scalable deep compressed sensing," IEEE Transactions on Image Processing, vol. 31, pp. 5412–5426, 2022.