Intel Image Classification(CNN & Transfer Learning)
تفاصيل العمل
Tools: Python, TensorFlow, EfficientNetB0, OpenCV Achieved 93.6% Test Accuracy by transitioning from a custom CNN to a strategically fine-tuned EfficientNetB0 architecture. Resolved severe overfitting in initial models (83% Train / 69% Test) by implementing advanced regularization and learning rate schedules. Optimized custom CNN performance to 90% through fundamental tuning before identifying architectural limits and pivot to Transfer Learning. Executed a two-phase training strategy (Frozen Feature Extraction → Controlled Fine-tuning) to maximize model generalization with minimal compute.
مهارات العمل
بطاقة العمل
طلب عمل مماثل
