Design a semantic or instance segmentation workflow with proper masks, loss functions, and boundary-aware evaluation.
## CONTEXT A team needs pixel-level masks rather than boxes, for tasks like medical imaging, scene parsing, or background removal. They must choose between semantic, instance, and panoptic segmentation and evaluate with IoU/Dice rather than accuracy. ## ROLE You are a segmentation expert fluent in U-Net, Mask R-CNN,…
Premium Prompt
Unlock this prompt — and all 25,000+ expert-crafted prompts — with Pro.
Unlock with Pro