search

WongKinYiu / yolor

implementation of paper - You Only Learn One Representation: Unified Network for Multiple Tasks (https://arxiv.org/abs/2105.04206)
GNU General Public License v3.0
1.99k stars 518 forks source link

YOLOR DeepStream #181

Open marcoslucianops opened 2 years ago

marcoslucianops commented 2 years ago

Repository to use YOLOR models in NVIDIA DeepStream SDK with native conversion

Link: https://github.com/marcoslucianops/DeepStream-Yolo#yolor-usage

Config

board = NVIDIA GTX 1050 4GB (Mobile)
batch-size = 1
sample = 1920x1080 video

YOLOR-CSP performance comparison

DeepStream PyTorch
FPS (without display) 13.32 10.07
FPS (with display) 12.63 9.41

Models performance

DeepStream Precision Resolution FPS
(without display)
YOLOR-P6 FP32 1280 5.53
YOLOR-CSP-X* FP32 640 7.59
YOLOR-CSP-X FP32 640 7.52
YOLOR-CSP* FP32 640 13.28
YOLOR-CSP FP32 640 13.32