YOLO-NAS

yolonasIn LibreYOLO

Deci AI (acquired by NVIDIA) · Deci AI technical report 2023

Paper Code

Variants3

Parameters19.0M - 67.0M

Best Paper mAP52.22%

LicenseApache-2.0

Model Variants

All variants benchmarked on COCO val2017. Click any variant for the full hardware breakdown.

Model	Params	GFLOPs	Paper mAP	Best FPS
YOLO-NAS-S	19.0M	32.8	47.5%	17 FPS NVIDIA A100 · PyTorch FP32	Details
YOLO-NAS-M	51.1M	88.9	51.55%	17 FPS NVIDIA A100 · PyTorch FP32	Details
YOLO-NAS-L	67.0M	116.6	52.22%	17 FPS NVIDIA A100 · PyTorch FP32	Details

Architecture

Type

one-stage

Backbone

Neural-Architecture-Search (AutoNAC)

Neck

PANet

Head

Decoupled (anchor-free, distribution focal loss)

anchor-freenmsTrained on Objects365 + COCO + Roboflow-100

Run any model with one line

LibreYOLO has the best catalogue of state-of-the-art detectors, all behind one MIT-licensed Python API.

from libreyolo import LibreYOLO, SAMPLE_IMAGE # LibreYOLO has the best catalogue of state-of-the-art models. model = LibreYOLO("LibreRFDETRl.pt") # RF-DETR-L (transformer flagship) results = model(SAMPLE_IMAGE, save=True) # run inference, save the annotated image # Swap in any other model, same one-line API (weights auto-download): # LibreYOLO("LibreYOLO9c.pt") # YOLO9-C # LibreYOLO("LibreYOLOXx.pt") # YOLOX-X # LibreYOLO("LibreDFINEx.pt") # D-FINE-X # LibreYOLO("LibreRTDETRr50.pt") # RT-DETR-R50