Table 3 Summarizes the key training hyperparameters and architectural settings for each model.

From: Robustness analysis of YOLO and faster R-CNN for object detection in realistic weather scenarios with noise augmentation

Model

Input size

Batch size

Learning rate

Optimizer

Weight decay

Epochs

LR scheduler

Loss function

YOLOv5s

640 × 640

16

0.001

Adam

0.0005

50

CosineAnnealing

CIoU Loss

YOLOv8m

800 × 800

16

0.001

Adam

0.0005

50

CosineAnnealing

CIoU Loss

YOLOv10n

384 × 384

16

0.001

Adam

0.0005

50

CosineAnnealing

CIoU Loss

Faster R-CNN

800 × 800

16

0.001

Adam

0.0001

50

StepLR (step = 10, γ = 0.1)

Cross-Entropy + Smooth L1

Back to article page