Code and results for one-class autoencoder verification of historical sketches using handcrafted features (Fourier energy, Shannon entropy, contrast, GLCM homogeneity, fractal dimension) and deep feature baselines (ResNet50, EfficientNet-V2). Companion repository to Ugail et al., PLOS ONE (2026).
computer-vision autoencoder digital-humanities glcm fourier-transform anomaly-detection one-class-learning fractal-dimension texture-analysis shannon-entropy handcrafted-features image-feature-extraction computational-art-history sketch-verification art-authentication historical-sketches
-
Updated
Apr 26, 2026 - Jupyter Notebook