• Semantic Segmentation: Focuses on classifying pixels into different classes (e.g., background vs. object). However, when objects touch or overlap, this method may classify them as a single contiguous mass.
• Instance Segmentation: Separates touching or overlapping objects into distinct entities, enabling more precise analysis especially in the case of touching or overlapping objects.