init git and download artgan files

.editorconfig

@@ -0,0 +1,12 @@
+# EditorConfig is awesome: https://EditorConfig.org
+
+# top-most EditorConfig file
+root = true
+
+[*]
+indent_style = space
+indent_size = 4
+end_of_line = lf
+charset = utf-8
+trim_trailing_whitespace = false
+insert_final_newline = false

.gitignore

@@ -0,0 +1,31 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+.Python
+venv/
+.venv/
+.env
+*.egg-info/
+.eggs/
+dist/
+build/
+
+# Data & outputs — large dirs and model files are not committed
+data/arty_cache/
+data/images/
+data/wikiart/
+data/wikiart.zip
+*.pth
+*.pt
+checkpoints/
+logs/
+runs/
+
+# Keep the metadata index in the repo (small, useful to commit)
+!data/wikiart_index.csv
+
+# IDE & OS
+.idea/
+.vscode/
+.DS_Store

.tool-versions

@@ -0,0 +1 @@
+python 3.13.2

data/artgan_csv/artist_class.txt

@@ -0,0 +1,23 @@
+0 Albrecht_Durer
+1 Boris_Kustodiev
+2 Camille_Pissarro
+3 Childe_Hassam
+4 Claude_Monet
+5 Edgar_Degas
+6 Eugene_Boudin
+7 Gustave_Dore
+8 Ilya_Repin
+9 Ivan_Aivazovsky
+10 Ivan_Shishkin
+11 John_Singer_Sargent
+12 Marc_Chagall
+13 Martiros_Saryan
+14 Nicholas_Roerich
+15 Pablo_Picasso
+16 Paul_Cezanne
+17 Pierre_Auguste_Renoir
+18 Pyotr_Konchalovsky
+19 Raphael_Kirchner
+20 Rembrandt
+21 Salvador_Dali
+22 Vincent_van_Gogh

data/artgan_csv/genre_class.txt

@@ -0,0 +1,10 @@
+0 abstract_painting
+1 cityscape
+2 genre_painting
+3 illustration
+4 landscape
+5 nude_painting
+6 portrait
+7 religious_painting
+8 sketch_and_study
+9 still_life

data/artgan_csv/style_class.txt

@@ -0,0 +1,27 @@
+0 Abstract_Expressionism
+1 Action_painting
+2 Analytical_Cubism
+3 Art_Nouveau_Modern
+4 Baroque
+5 Color_Field_Painting
+6 Contemporary_Realism
+7 Cubism
+8 Early_Renaissance
+9 Expressionism
+10 Fauvism
+11 High_Renaissance
+12 Impressionism
+13 Mannerism_Late_Renaissance
+14 Minimalism
+15 Naive_Art_Primitivism
+16 New_Realism
+17 Northern_Renaissance
+18 Pointillism
+19 Pop_Art
+20 Post_Impressionism
+21 Realism
+22 Rococo
+23 Romanticism
+24 Symbolism
+25 Synthetic_Cubism
+26 Ukiyo_e

doc/data_preparation.md

@@ -0,0 +1,21 @@
+# ArtGAN WikiArt
+
+We use WikiArt dataset which has the source images and the indexes for each of the 3 categories already classified.
+We need to download the images dataset and generate `data/wikiart_index.csv`.
+
+## 1. Images
+
+The refined WikiArt dataset (images only) is provided by the [ArtGAN project](https://github.com/cs-chan/ArtGAN/tree/master/WikiArt%20Dataset).
+
+- **URL:** [wikiart.zip on Google Drive](https://drive.google.com/file/d/1vTChp3nU5GQeLkPwotrybpUGUXj12BTK/view) (~25.4 GB).
+- **License:** Non-commercial research only; see [ArtGAN README](https://github.com/cs-chan/ArtGAN/blob/master/WikiArt%20Dataset/README.md).
+
+We download the file into the repo, extract it under `data/wikiart`. The zip and the extracted files are excluded from git. `rembrandt_woman-standing-with-raised-hands.jpg` and `vincent-van-gogh_l-arlesienne-portrait-of-madame-ginoux-1890.jpg` contained errors.
+
+The root data dir `data/wikiart/` contains 27 style subdirectories, each containing images named `{artist-slug}_{painting-title}-{year}.jpg`.
+
+## 2. ArtGAN label files for metadata
+
+ArtGAN only uses a subset of WikiArt, ArtGAN’s artist, style and genre labels [wikiart_csv.zip] (https://drive.google.com/file/d/1uug57zp13wJDwb2nuHOQfR2Odr0hh1a8/view)
+which we include in the repo for convenience `data/artgan_csv`. There's a slight variation from in the style `Art_Nouveau` to `Art_Nouveau_Modern`.
+Using `scripts/validate_artgan_paths.py` to validate

doc/setup.md

@@ -0,0 +1,21 @@
+# Development setup
+
+## Runtimes (asdf)
+
+The project uses [asdf](https://asdf-vm.com/) for runtimes. Versions are pinned in [`.tool-versions`](../.tool-versions).
+
+```bash
+# Install the plugin and runtime if needed
+asdf plugin add python
+asdf install   # installs versions listed in .tool-versions
+```
+
+## Python env
+
+Created a virtualenv and install dependencies:
+
+```bash
+python -m venv .venv
+. .venv/bin/activate 
+pip install -r requirements.txt
+```

scripts/validate_artgan_paths.py

@@ -0,0 +1,62 @@
+"""
+scripts/validate_artgan_paths.py
+=================================
+Check that every path in the ArtGAN train/val CSVs exists under data/wikiart/.
+"""
+import csv
+import sys
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parent.parent
+WIKIART_ROOT = REPO_ROOT / "data" / "wikiart"
+CSV_DIR = REPO_ROOT / "data" / "artgan_csv"
+
+CSV_FILES = [
+    "style_train.csv",
+    "style_val.csv",
+    "genre_train.csv",
+    "genre_val.csv",
+    "artist_train.csv",
+    "artist_val.csv",
+]
+
+def main() -> None:
+    if not WIKIART_ROOT.exists():
+        print(f"ERROR: {WIKIART_ROOT} not found.")
+        sys.exit(1)
+
+    total_files = 0
+    all_ok = True
+    for csv_file_name in CSV_FILES:
+        csv_file_path = CSV_DIR / csv_file_name
+        label = csv_file_name.replace(".csv", "")
+        if not csv_file_path.exists():
+            print(f"  {label}: SKIP (file not found)")
+            continue
+
+        missing = []
+        total = 0
+        with open(csv_file_path) as csv_file:
+            for row in csv.reader(csv_file):
+                if not row:
+                    continue
+                total += 1
+                local_path = row[0].strip()
+                if not (WIKIART_ROOT / local_path).exists():
+                    missing.append(local_path)
+        if missing:
+            all_ok = False
+            print(f"  {label}: {total} rows, MISSING {len(missing)} files")
+            for p in missing[:5]:
+                print(f"    - {p}")
+            if len(missing) > 5:
+                print(f"    ... and {len(missing) - 5} more")
+        else:
+            print(f"  {label}: {total} rows, OK")
+        total_files += total
+    if not all_ok:
+        sys.exit(1)
+    print(f"{total_files} total files, OK")
+
+if __name__ == "__main__":
+    main()