Russian PhD Student Builds AI-Powered ‘X-Ray’ Vision System
The neural network can reconstruct hidden details from ordinary smartphone images — a breakthrough for medicine and robotics.
A postgraduate researcher from Samara National Research University has developed a neural network–based imaging system that acts like an AI-powered X-ray, allowing computers to detect and model hidden details of objects using a simple RGB camera instead of expensive laser scanners.
The project, led by Gennady Algashov, an engineer and PhD student at the university’s Laboratory of Automated Research Systems, uses artificial intelligence to determine an object’s key points and shape in real time.
Teaching AI the “Language of Points”
To train the neural network, Algashov created a synthetic image generator that automatically “photographs” 3D models from multiple angles and lighting conditions, adding noise to simulate real-world scenarios. The system labels object boundaries and key points automatically, allowing rapid training on any type of model.
The result is a real-time neural vision system that can identify the position and orientation of objects — even when they’re partially hidden. Researchers believe the technology could transform robotics, medical imaging, and industrial diagnostics, offering fast, affordable alternatives to traditional 3D scanning systems.
