Page 72 - eBook_Proceedings of the International Conference on Digital Manufacturing V2
P. 72
Proceedings of the International Conference on Digital Manufacturing –
Volume 2
Figure 24: Confusion matrix showing model predictions across
foreground cervical cell classes
Figure 25: Recall scores for each foreground cervical cell classes
Despite high recall, the model exhibited lower F1-scores
across classes, primarily due to decreased precision. Lower
precision indicates a higher frequency of false-positive detections,
suggesting that the model occasionally misclassified pixels,
predicting them as belonging to a particular class when they did
not.
56
