Add: Dataset validation and docs

.gitattributes

@@ -52,3 +52,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.jpg filter=lfs diff=lfs merge=lfs -text
 *.jpeg filter=lfs diff=lfs merge=lfs -text
 *.webp filter=lfs diff=lfs merge=lfs -text
+# Unum files - uncomperessed
+*.usearch filter=lfs diff=lfs merge=lfs -text
+*.fbin filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,2 @@
+images/*
+.DS_Store

README.md

@@ -1,3 +1,25 @@
 ---
 license: apache-2.0
 ---
+
+# 25K Unsplash Images for Search
+
+This is a derivative work based on two existing datasets.
+
+- `images.csv` metadata from [Unsplash](https://github.com/unsplash/datasets), sorted and converted to CSV.
+- `images/` in 250x250 resolution by [kaggle/@jettchentt](https://www.kaggle.com/datasets/jettchentt/unsplash-dataset-images-downloaded-250x250).
+- `images.fbin` is a binary file with UForm image embeddings.
+- `images.usearch` is a binary file with a serialized USearch index.
+
+The original `images.tsv` from Unsplash has been filtered to avoid missing images.
+The embeddings and the index can be reconstructed with the `main.py` script.
+On the Apple M2 Pro CPU:
+
+- Image vectorization takes 100ms/image, or 10 inferences/second.
+- Indexing vectors one-by-one happens at 700 vectors/second speed.
+
+To rebuild the indexes:
+
+```sh
+./main.py
+```

main.py

@@ -1,4 +1,5 @@
-from os import listdir, path, PathLike
+#!/usr/bin/env python3
+from os import listdir, path, PathLike, remove
 from os.path import isfile, join
 
 import pandas as pd
@@ -24,9 +25,13 @@ def is_image(path: PathLike) -> bool:
         return False
 
 
+def trim_extension(filename: str) -> str:
+    return filename.rsplit('.', 1)[0]
+
+
 names = sorted(f for f in listdir('images') if is_image(join('images', f)))
+names = [trim_extension(f) for f in names]
 
-names = [filename.rsplit('.', 1)[0] for filename in names]
 table = pd.read_table('images.tsv') if path.exists(
     'images.tsv') else pd.read_table('images.csv')
 table = table[table['photo_id'].isin(names)]
@@ -35,6 +40,13 @@ table.reset_index()
 table.to_csv('images.csv', index=False)
 
 names = list(set(table['photo_id']).intersection(names))
+names_to_delete = [f for f in listdir(
+    'images') if trim_extension(f) not in names]
+
+if len(names_to_delete) > 0:
+    print(f'Plans to delete: {len(names_to_delete)} images without metadata')
+    for name in names_to_delete:
+        remove(join('images', name))
 
 model = get_model('unum-cloud/uform-vl-english')
 vectors = []