Production-ready time-series image classifier using EfficientNet/NFNet with PyTorch Lightning. Supports single-modal and multi-modal learning with F1-score optimization.

QV-Pipe multi-label classification (super-image 3×3, timm, ASL/CB-Focal, OneCycle, EMA, 5-fold ensemble).. Inspired by Challenge Pipe 2022.

🔍 Classify defects in Quick-View videos using deep learning with a complete pipeline for data extraction, model training, and reproducibility.