🚀 5 Mistakes I Made in My First CNN Project (That Ruined My Results)

The author shares five key mistakes from their first CNN project on brain tumor MRI classification: ignoring class imbalance, unnecessarily increasing model complexity, relying on overall accuracy instead of per-class metrics, blindly copying hyperparameters, and skipping early data visualization. These errors led to biased predictions, overfitting, and confusion between visually similar tumor types. The core lesson is that multi-class classification demands careful data understanding and per-class evaluation beyond simple accuracy.