Cignal Powers Newly Launched DHS S&T Synthetic Baggage Image Dataset to Accelerate Airport Screening Innovation

Following today’s announcement by the Department of Homeland Security (DHS) Science and Technology Directorate (S&T), Cignal LLC is proud to highlight its foundational role in building the newly launched synthetic baggage image dataset, a critical milestone designed to accelerate the next generation of airport screening technology.

As detailed in today’s DHS S&T press release, the new dataset provides Original Equipment Manufacturers (OEMs) and third-party AI developers with unprecedented access to high-fidelity, physically accurate X-ray and CT synthetic images. This resource empowers developers to build, test, and validate automatic threat recognition (ATR) algorithms against complex scenarios without the bottlenecks of physical data collection.

"The AI systems protecting our checkpoints require rigorous, dynamic environments," said Jaclyn Fiterman, CEO of Cignal. "We are thrilled to support DHS S&T in releasing this dataset. By providing developers with access to high-quality synthetic scenes, we are ensuring that highly adaptive threat detection models can be deployed faster, securely, and with greater accuracy."

Building What Doesn't Exist Yet
The capability to produce this extensive dataset is the direct result of Cignal’s multi-phase collaboration with DHS through the Silicon Valley Innovation Program (SVIP). Cignal serves as an agent for high-stakes vision AI. Generating physically accurate, fully labeled scenes - such as those found in this new dataset - is just one way Cignal allows defense and security teams to build what they need to see.

Leveraging patented voxel-tensor physics and a rendering engine, Cignal’s technology allows for the creation of X-ray, CT, millimeter-wave, and multi-spectral environments from natural language prompts. This means developers can prototype and red-team visual AI against threats that don't exist yet or defects that are too infrequent to capture in the real world.

Advancing the Future of Vision AI
While generating synthetic data solves the "cold start" and "long tail" problems in model training, Cignal’s vision agents go further. Cignal allows operators to iterate on those scenes, asking what's in them and discovering what works and what doesn't. This includes Cignal Ratio, the first compositional Vision Language Model (VLM) in security screening, providing AI that shows its work and delivers vital visual decision support for officers and first responders.

This DHS dataset release represents a major step forward in public-private partnerships, equipping researchers and developers with the tools to tackle the hardest detection challenges.

To read the official release from the DHS Science and Technology Directorate, please click here.

To learn more about Cignal and how we act as your agent for high-stakes vision AI, visit cignalai.com.

‍