Skip to main content
. 2024 Apr 18;15:1373590. doi: 10.3389/fpls.2024.1373590

Table 3.

Default hyperparameters for training.

Hyperparameter Description Default Value
epochs Number of epochs 200
patience Epochs to wait for early stopping 50
batch Batch size 32
workers Concurrent Threads 16
imgsz Resize image to 256x256 pixels
pretrained Use Pre-Trained weights True
optimizer Optimizer AdamW
freeze Disable weight update for these layers None
lr0 Initial Learning rate 0.001429
momentum Learning Rate momentum 0.9
box Box loss gain 7.5
cls Class loss gain 0.5
dfl DFL loss gain 1.5