Figure 14. Ablation study conducted on HierbaNetV1 with three blocks to comprehend each block’s contribution: feature maps generated by (A) Layer 1-Base_Conv1 (Conv2D), (B) Layer 25-B1_LL_Conv1 (Conv2D), (C) Layer 33-Base_Conv2 (Conv2D), (D) Layer 57-B2_LL_Conv1 (Conv2D), (E) Layer 65-Base_Conv3 (Conv2D), (F) Layer 89-B3_LL_Conv1 (Conv2D), and (G) Layer 97-Base_Conv4 (Conv2D).
