AI Deep Learning Gets Upgrade With New Object Detection Model

An Israeli company that uses artificial intelligence to develop artificial intelligence has unveiled a new deep learning model to detect objects in real time, which it claims outstrips its predecessors. 

Ramat Gan-based Deci says that its YOLO-NAS model, which was also created through AI, improves on previous versions that struggled to accurately process certain data. According to Deci, YOLO-NAS can process more data at a faster pace than its predecessors. 

“The release of YOLO-NAS is a major leap forward for inference performance and efficiency of object detection models, addressing the limitations of previous YOLO models and offering unprecedented adaptability for diverse tasks and hardware,” says Deci CEO Yonatan Geifman.

Deep learning involves teaching computers to mimic the layered processes of the human mind, for example allowing driverless cars to distinguish between people and inanimate objects in its environment. 

According to the company, YOLO-NAS “pushes the boundaries of object detection with superior real-time object detection capabilities.”

It was created using Deci’s Neural Architecture Search Technology AutoNAC, which allows users to build their own deep learning models.

Prof. Ran El-Yaniv, the company’s chief scientist, said: “The devil is in the details, and designing such optimal architectures is an incredibly complex task, often too difficult for humans to tackle alone.

“Deci’s pioneering AutoNAC engine empowers AI teams to construct state-of-the-art architectures that impeccably align with their applications, delivering unparalleled results.”

The company says its mission is to provide tools for AI developers to innovate and create AI-based solutions. It has made YOLO-NAS open source and available for non-commercial use through its SuperGradients computer vision training library. 

Deci is inviting users to access YOLO-NAS and its other deep learning models available via SuperGradients and offer feedback on their experiences.