Add dataset card

README.md

@@ -1,5 +1,37 @@
-This is a dataset dervied from placing watermarks on the CalebA dataset of huma faces. The purpose of this dataset is to train a model to be able to remove watermarks from images
-
 ---
-license: mit
+dataset_info:
+  features:
+    - name: pair_id
+      dtype: int64
+    - name: clean_image
+      dtype: image
+    - name: watermarked_image
+      dtype: image
+  splits:
+    - name: train
+      num_examples: 5000
 ---
+
+# Watermark Dataset
+
+Paired clean / watermarked face images generated from CelebA.
+
+Each row contains:
+- `pair_id` — unique integer index
+- `clean_image` — original 128×128 face crop
+- `watermarked_image` — same image with a synthetic proof watermark overlay
+
+## Usage
+
+```python
+from datasets import load_dataset
+
+# Streaming
+ds = load_dataset("GeraldNdawula/Watermark_Dataset", streaming=True, split="train")
+for sample in ds:
+    clean       = sample["clean_image"]         # PIL Image
+    watermarked = sample["watermarked_image"]   # PIL Image
+
+# Full load
+ds = load_dataset("GeraldNdawula/Watermark_Dataset", split="train")
+```