Software

In the spirit of reproducible research, the following toolboxes and datasets serve as companions of several publications.

Python Toolboxes

• HyperBench: A benchmarking framework for fusion-based hyperspectral super-resolution methods. HyperBench provides a standardized and reproducible pipeline for generating synthetic degradations, evaluating reconstruction quality, and logging structured results. (Developed by Ritik Shah)

• SpectraMorph: A GitHub repository that implements a lightweight, interpretable, and self-supervised framework for hyperspectral and multispectral image fusion. Also includes scripts and data to generate figures from the preprint. (Developed by Ritik Shah)

• SpectraLift: A GitHub repository that implements a lightweight self-supervised framework for hyperspectral and multispectral image fusion. Also includes scripts and data to generate figures from the preprint. (Developed by Ritik Shah)

Matlab Toolboxes

• Few-Shot Learning-Based Human Activity Recognition: A GitHub repository that implements parameter transfer-based few-shot learning scheme for human activity recognition from wearable sensor data. (Developed by Siwei Feng)

• Graph Autoencoder-Based Unsupervised Feature Selection: A Matlab toolbox that implements a feature selection algorithm that enforces a neighborhood graph regularization within the latent space of an autoencoder model for the data. (Developed by Siwei Feng)

• Crowd-Resilient Visual Localization Algorithm: A Matlab toolbox that implements a crowd-resilient localization algorithm based on images. Also includes scripts and data to generate figures from the conference paper. (Developed by Zhuorui Yang)

• Hyperspectral Band Selection from Statistical Wavelet Models: A Matlab toolbox that implements a band selection algorithm for hyperspectral imaging based on a Hidden Markov Tree model for the wavelet coefficients of the spectra. (Developed by Siwei Feng)

• Manifold-Aware Pixel Selection (MAPS): A Matlab toolbox that implements mask design algorithms for manifold-modeled image datasets. Also includes scripts and data to generate figures from the journal paper. (Developed by Hamid Dadkhahi)

• Manifold Analysis GUI (MAGI): A Matlab toolbox that allows for intuitive visualization and navigation of image manifolds. (Developed by Kevin Eykholt)

• Compressive Parameter Estimation via Polar Interpolation: Includes algorithm implementations for the IBOMP algorithm and applications to time-delay estimation from compressive measurements. Also includes scripts and data to generate figures from the journal paper. (Developed by Karsten Fyhn)

• Spectral Compressive Sensing via Polar Interpolation: Includes algorithm implementations for BISP and CBP. Also includes scripts to generate figures from the conference paper. (Developed by Karsten Fyhn and Hamid Dadkhahi)

• Universal Compressed Sensing Estimation: Includes algorithm implementations for MAP estimation of compressively sensed signals using empirical universal priors. Also includes scripts to generate figures from the conference paper.

• Spectral Compressive Sensing: Includes algorithm implementations for spectral compressive sensing recovery of frequency-sparse signals. Also includes scripts and data to generate figures from the journal paper.

• Kronecker Compressive Sensing: This toolbox contains algorithms and data to generate figures from the journal paper.

  • Full toolbox - 3.5GB, contains scripts and data files to generate paper figures.

  • Basic toobox - 100MB, contains scripts to generate data for paper figures.

• Model-Based Compressive Sensing: Includes algorithm implementations for model-based compressive sensing recovery for a variety of structured sparsity models (Developed by Chinmay Hegde).

Datasets

• Mug and Koalas Joint Manifolds Datasets: Provides data from multi-camera acquisition of low-dimensional object articulations. These datasets were used for joint manifold learning in the joint manifolds journal paper (Compiled by Chinmay Hegde).

• Single Pixel Camera Dataset: Includes measurement data and recovery scripts for several test single-pixel camera compressive images, as detailed in the magazine article.

• Acoustic Vehicle Classification Dataset: Includes raw dataset and feature extraction algorithms for acoustic vehicle dataseries from the SITEX02 experiment in Twenty-Nine Palms, CA, as detailed in the journal paper.

  • Timeseries files: This ZIP file contains the original binary timeseries for all runs from the SITEX02 experiment, November 2001, Twentynine Palms, CA.

  • Energy files: This ZIP file contains the energy series for all runs from the SITEX02 experiment.

  • Event Files: This ZIP file contains the event timeseries and feature files for all runs from the SITEX02 experiment. The datasets have been converted to Matlab format by Erdem Köse of Gebze Technical University and are available for download here