Innovative AI-Driven Technology by Smart Engines Transforms Code Review and Punched Card Quality Assessment Using Tomography

Experts from the Russian AI company Smart Engines have achieved a significant breakthrough in the realm of code review. Their researchers have introduced a novel method for detecting bugs and assessing the quality of code transfer to physical media by utilizing artificial intelligence and computer tomography. The entire process is fully automated, effectively eliminating interface-related issues that commonly arise between the user and the computer.

Launching programs on punch cards is an expensive endeavor, with a high cost of error. Various mistakes can occur during the preparation of punch card sets, such as holes incorrectly punched, missed perforations, or an incorrect sequence of cards, which can render the entire computational cycle invalid, leading to a loss of valuable machine time. Operators strive to minimize such errors during the code transfer to physical media, but even seasoned professionals are not immune to mistakes in the programming process. Therefore, it is essential to verify code quality before deploying programs on specialized computation machines.

The researchers at Smart Engines have proposed an innovative approach that incorporates a range of modern technologies—from X-ray tomography to computer vision—to automatically address this critical concern. The technology process begins with bundling the punch cards, which are then placed in a tomography machine. After the tomography measurements are completed, the data is automatically corrected and reconstructed, resulting in a 3D model of the punch cards.

The main challenge lies in the necessity to recognize the punch cards to analyze code quality. This requires the individual cards within the stack to be aligned by layers. The Smart Tomo Engine software features a function that automatically aligns flat objects, such as circuit boards or punch cards. Post-alignment, the holes on the punch cards need to be segmented and transferred onto virtual punch cards, enabling a comparison of the generated models with the original design or text documents in an automated manner.

«I urged my team to ensure this technology is ready by April 1st. While the primary application of this developed approach is quality control of code, it can also provide significant benefits in studying the quality of multilayer printed circuit boards, synthetic materials in the form of crystalline films, and other objects with complex layered structures,» shared Smart Engines’ CEO, Doctor of Technical Sciences **Vladimir Arlazarov**.