Supplementary Artifacts for Gated Multi‑Scale Feature Fusion: Preprocessed SIPaKMeD Cervical Cell Image Dataset, Extracted Features, Model Weights, and Classification Results on SIPaKMeD (5‑Fold Cross‑Validation)

This archive contains all experimental artifacts produced during the training and evaluation of the gated multi‑scale feature fusion network described in our manuscript (under review at *The Visual Computer*). It includes: 1. **Preprocessed SIPaKMeD dataset** (5‑fold split, resized to 224×224) – the same as the dataset version.2. **Extracted features** from DenseNet121 and RepLKNet31B backbones (saved as `.npy` files in `features/`).3. **Trained model checkpoints** for the gated fusion classifier (`.pth` files) and standalone baselines.4. **Classification results** – JSON files with accuracy, F1, confusion matrices, and per‑fold logs.5. **Figures** – loss curves, accuracy plots, and confusion matrices. This package enables full reproduction of our reported results without re‑running feature extraction or retraining. **Original dataset citation:** Plissiti et al., ICIP 2018 (DOI: 10.1109/ICIP.2018.8451788). **Related software repository:** https://github.com/liuyue319/Gated-MultiScale-CervicalCell (DOI: 10.5281/zenodo.20291083)