Figure 4
The figure contains six confusion matrix heatmaps arranged in two rows of three panels, labeled “(a) Fold 1”, “(b) Fold 2”, “(c) Fold 3”, “(d) Fold 4”, “(e) Fold 5”, and “(f) Test set”. Each heatmap has the horizontal axis labeled “Predicted” and the vertical axis labeled “Actual”. Both axes list six classes: “Chickenpox”, “Cowpox”, “H F M D”, “Healthy”, “Measles”, and “Monkeypox”, from top to bottom on the vertical axis and from left to right on the horizontal axis. Each matrix includes a vertical color bar at the right indicating count intensity, where darker shades represent higher values. The color bars for five folds range from 0 to 300 with an interval of 50, while for the test set, it ranges from 0 to 35 with an interval of 5. The data from the rows for each heatmap are as follows: Fold 1 values by row are Actual Chickenpox: 86, 0, 0, 0, 0, 0; Actual Cowpox: 0, 72, 0, 0, 0, 3; Actual H F M D: 0, 0, 183, 0, 0, 1; Actual Healthy: 1, 0, 0, 128, 0, 1; Actual Measles: 1, 0, 0, 1, 60, 1; Actual Monkeypox: 0, 1, 0, 0, 0, 324. Fold 2 values are Chickenpox: 84, 0, 0, 0, 0, 2; Cowpox: 0, 72, 0, 0, 0, 3; H F M D: 0, 0, 184, 0, 0, 0; Healthy: 0, 0, 0, 128, 0, 2; Measles: 0, 0, 0, 0, 61, 1; Monkeypox: 2, 0, 3, 0, 0, 321. Fold 3 values are Chickenpox: 84, 0, 0, 0, 0, 1; Cowpox: 2, 72, 1, 0, 0, 1; H F M D: 0, 0, 182, 0, 0, 2; Healthy: 1, 0, 0, 126, 0, 3; Measles: 0, 0, 0, 0, 60, 2; Monkeypox: 4, 0, 1, 0, 0, 321. Fold 4 values are Chickenpox: 82, 0, 1, 0, 0, 2; Cowpox: 0, 73, 0, 0, 0, 3; H F M D: 0, 0, 183, 0, 0, 1; Healthy: 0, 0, 0, 127, 1, 2; Measles: 0, 0, 0, 0, 62, 0; Monkeypox: 3, 0, 2, 0, 1, 319. Fold 5 values are Chickenpox: 82, 0, 0, 1, 0, 2; Cowpox: 1, 73, 1, 0, 0, 1; H F M D: 0, 0, 181, 1, 0, 2; Healthy: 0, 0, 2, 125, 1, 1; Measles: 0, 0, 2, 0, 61, 0; Monkeypox: 0, 0, 2, 0, 0, 323. The test set matrix values are Chickenpox: 9, 0, 1, 0, 0, 2; Cowpox: 1, 5, 0, 0, 0, 4; H F M D: 0, 0, 22, 1, 0, 2; Healthy: 0, 0, 1, 15, 0, 2; Measles: 0, 0, 0, 0, 6, 3; Monkeypox: 2, 0, 1, 1, 0, 39.Confusion matrices of the proposed model over five folds and the test set on the MSLD v2 dataset, demonstrating robust and consistent performance