Figure 3. Architecture of HierbaNetV1_Block I: Block I and Block II define the novelty and key characteristics of our proposed model.

Each block follows three crucial steps: Firstly, the four diversified filters with kernel sizes $1\times 1$, $3\times 3$, $5\times 5$, and $7\times 7$ extract high-level features $(F_{HL})$. Secondly, low-level features $(F_{LL})$ are extracted by filters of kernel sizes $3\times 3$. Thirdly, $F_{HL}$ and $F_{LL}$ are integrated to form F thus producing a rich set of features with multiple levels of complexity.