Fix README metadata and formatting for Hugging Face

README.md

@@ -1,4 +1,3 @@
-
 ---
 annotations_creators:
 - human
@@ -13,21 +12,29 @@ source_datasets:
 task_categories:
 - image-segmentation
 task_ids:
-- multi-class-segmentation
+- semantic-segmentation
 ---
 
-
-This dataset contains user interface (UI) images along with associated prompts and segmentation outputs...
 # VisionReasoner UI Dataset
 
 This dataset contains user interface (UI) images along with associated annotation prompts and solutions for fine-tuning the VisionReasoner model.
 
 ## Structure
 
-- `images/`: Folder containing UI images (.webp).
-- `visionreasoner_dataset.parquet`: A `.parquet` file with metadata like problem descriptions, image paths, and solutions.
-- `demo.py`: Hugging Face dataset loading script.
+- `images/`: Folder containing UI images (`.webp` format).
+- `visionreasoner_dataset.parquet`: A `.parquet` file with metadata such as:
+  - Prompt (`problem`)
+  - Segmentation solution (`solution`)
+  - Image file reference
+  - Image height
+- `demo.py`: A custom dataset loading script using Hugging Face `datasets` library.
+
+## Usage
+
+To load and preview the dataset:
 
-## Preview
+```python
+from datasets import load_dataset
 
-Use `datasets.load_dataset("shirve13/Demo", trust_remote_code=True)` to load this dataset.
+dataset = load_dataset("shirve13/Demo", trust_remote_code=True)
+print(dataset["train"][0])