Design, Test, & Deploy Next Generation X-Ray Systems with Quadridox & Rendered.ai

One secret to building performant computer vision algorithms faster is aligning with the right partners who can provide advanced domain expertise.  

Building a Community of Experts

Rendered.ai has cultivated a network of best-in-class simulation partners in various areas of vision-based AI and ML. Along with Rendered.ai’s team of synthetic data generation experts, these partners guide customers with innovative ways to reduce the time, expense, and frustration that can come with shaping computer vision systems for specific domains. The Rendered.ai platform further enables end customers to generate massive datasets that fill gaps in computer vision training data by repackaging domain-specific partner capabilities, like Quadridox’s QSim RT Engine, in scalable cloud instances. 

The expertise of the Rendered.ai team, its network of partners, and the efficiencies built into the Rendered.ai data engineering platform make a powerful combination for data scientists to customize simulation and 3D content to successfully emulate real-world imagery. 

The Rendered.ai & Quadridox Partnership 

Over the last three years, Rendered.ai has partnered with Quadridox for its domain expertise in X-ray simulation. Quadridox is the trusted provider of tools and data for designing, testing, and deploying next-generation X-ray systems and algorithms. Customers can tap into Quadridox’s QSim RT Engine in the Rendered.ai synthetic data generation platform for fast, accurate, and scalable X-ray simulation.   

Quadridox works closely with the Rendered.ai team to generate repeatable synthetic data generation workflows in the Rendered.ai PaaS, which are used to experiment with and validate simulations with new 3D objects, sensors, and scenes. This collaboration enables Quadridox to offer endless customization in X-ray simulation with a broader range of complex, relevant objects to train AI and ML models.  

The Quadridox QSim RT Engine 

Quadridox’s QSim RT is a rapid, end-to-end X-ray physics simulation package capable of modeling a variety of testing methodologies. This engine simulates X-ray detection capabilities for security screening systems, food inspection systems, and more, with a combination of speed and accuracy unmatched in the market. QSim RT simulates everything from raw 2D radiographs and sinograms to full multi-energy 3D CT bag volumes that can be injected into a customer’s pipeline at any point in their design, development, and testing chain. 

The Quadridox team works closely with regulatory agencies and OEMs to help establish methods for validating synthetic X-ray tools and data that businesses can rely on. The QSim RT framework has undergone evaluation via independent agencies and has successfully demonstrated its value in human observer studies and algorithm training/testing. 

Case Study: QSim RT & the Rendered.ai PaaS in Action 

Use Case

Rendered.ai and Quadridox worked directly with a security technology company to enable the development of threat detection algorithms before a new offering was publicly released. Using the Rendered.ai platform, QSim RT was deployed to create a configurable X-ray imagery simulation workflow.  

The Rendered.ai team created a bag-packing algorithm that simulated the randomized arrangement of items like glasses, watches, charging cables, handguns, rifles, knives, and explosives inside different bags on the X-ray machine’s conveyor belt. 

Results

Over 69,000 X-Ray Images Rapidly Generated 

With the collaborative expertise and solutions of Rendered.ai and Quadridox, the customer could create over 69,000 highly variable X-ray images in a fraction of the time it would typically take to collect images in real-world environments. This was all accomplished before the new system was even built! The synthetic imagery produced was validated using reference imagery from the customer’s lab. It was then used to train and calibrate the threat detection algorithms deployed in the final product. 

Data That Matters

Quadridox and Rendered.ai’s collaboration provides the only way to generate physically accurate synthetic X-ray data to train computer vision with customizable and repeatable workflows in a commercially accessible solution. Joint customers have already demonstrated the efficacy of using synthetic X-ray data to improve algorithm training successfully. These customers save significant time and money using simulated X-ray data with Quadridox QSim RT and the Rendered.ai platform to generate scalable synthetic datasets for the ongoing training, improvement, and updating of performant computer vision algorithms. 

Connect with one of our experts and see a live demo of accelerated synthetic X-ray data generation with Rendered.ai and Quadridox.