Table 3. Trained model results on the original images from the dataset and the images where different filtering techniques were applied and hair was removed.

Metrics	Filtering technique combined with hair removal
	None	Fourier	Wavelet	Average blur	Low-pass
Accuracy (AC)	91.10%	90.90%	91.30%	90.80%	90.00%
Recall (SE)	87.60%	88.20%	87.00%	87.40%	89.80%
Precision (PR)	94.19%	93.23%	95.19%	93.78%	90.16%
F1 score (F1 sc)	90.78%	90.65%	90.91%	90.48%	89.98%