Lexset Releases Synthetic SAR Dataset for Public Testing

Lexset.ai has made a synthetic Synthetic Aperture Radar dataset based on the MSTAR benchmark available for public download and testing. The release, developed with the Air Force Research Laboratory under a Phase II SBIR effort, is meant to give researchers more accessible training data for radar imaging work.

Why it matters: - The public release gives researchers and developers new synthetic SAR data for machine learning and radar imaging tests. - Accessible training data remains a bottleneck for specialized computer vision work. - The dataset is intended to support research applications without requiring scarce real-world radar collections.

What happened: - Lexset.ai released a synthetic Synthetic Aperture Radar dataset based on the MSTAR dataset for public testing and research. - The dataset was developed with the Air Force Research Laboratory as part of a Small Business Innovation Research Phase II contract. - The release was announced Sept. 15, 2026, in Brooklyn, New York. - The synthetic MSTAR dataset is available for public download and testing on Kaggle.

The details: - The project focused on synthetic data generation technology for Department of the Air Force needs. - The release is described as an advancement in accessible training data for machine learning applications in radar imaging. - Francis Bitonti, Lexset.ai CEO, said the company was built after experiencing the difficulty of finding high-quality training data for specialized computer vision applications. - Bitonti said the dataset and techniques from the program are a step toward giving researchers and developers accessible SAR data to advance their work. - The release includes a mandatory disclaimer that the views expressed do not necessarily reflect the official policy or position of the Department of the Air Force, the Department of Defense, or the U.S. government.

Between the lines: - The release fits a broader Pentagon push to move small-business technology faster from proposal to award and into use. - AFRL and AFWERX have worked to streamline SBIR and STTR processes by reducing bureaucratic overhead and expanding access for small businesses. - Lexset is positioning its synthetic-data tools as relevant to defense and broader computer vision needs, not just radar imaging.

What's next: - Researchers can now download the dataset and use it to test synthetic SAR methods and machine learning workflows. - The public release could help validate whether synthetic radar data can close gaps where real data is limited or sensitive. - Lexset is likely to use the release to showcase its synthetic data generation capability for future defense and commercial work.

The bottom line: - Lexset is turning a defense-backed synthetic data effort into a public research asset, with the potential to widen access to radar training data and speed up experimentation in SAR imaging.