Matrix_One is a high-performance Image Quality Assessment (IQA) model developed for competitive benchmarking.
- Lightweight deep learning architecture
- Custom hybrid loss (MSE + Pearson + Rank Loss)
- Optimized for leaderboard performance
- Strong validation performance
- Competitive Kaggle-style submission
matrix (1).ipynb→ Training & inference pipelinesubmission.csv→ Final predictionsMatrix_One.pdf→ Detailed project report
- Python
- PyTorch
- NumPy / Pandas
- Model ensembling
- Advanced augmentation
- Hyperparameter tuning
Sanjeev Ranjan