YOLO-NAS (You Only Look Once Neural Architecture Search) is a cutting-edge image detection model that achieves state-of-the-art (SOTA) performance while offering exceptional accuracy and speed. This model surpasses the capabilities of previous models such as YOLOv5, YOLOv6, YOLOv7, and YOLOv8.

The development of YOLO-NAS is attributed to Deci&#39;s revolutionary Neural Architecture Search technology, known as AutoNAC&trade;. By leveraging the power of AutoNAC&trade;, users can input their specific task requirements, data characteristics (without needing access to the data itself), inference environment, and performance targets. The engine then guides users in discovering the optimal architecture that strikes the perfect balance between accuracy and inference speed, tailored to their unique application. Additionally, AutoNAC&trade; takes into account other crucial components in the inference stack, including compilers and quantization, ensuring comprehensive optimization.

In terms of performance metrics, YOLO-NAS demonstrates approximately 0.5 mean Average Precision (mAP) point higher accuracy compared to equivalent versions of YOLOv8 and YOLOv7. Moreover, it delivers a significant speed improvement of 10-20% when compared to these previous iterations.

The integration of YOLO-NAS in image detection tasks offers researchers and practitioners a powerful tool that combines outstanding accuracy and impressive speed, making it an ideal choice for various applications requiring real-time object detection and recognition.

Yolo-Nas

input_text

detected_image

YOLO-NAS is a groundbreaking object detection foundational model. It is the product of advanced Neural Architecture Search technology, meticulously designed to address the limitations of previous YOLO models. YOLO-NAS represents a major leap in object detection.

Yolo-Nas

About Model

Marketplace

Company

Legal

Follow Us