Update README.md

README.md

@@ -29,3 +29,74 @@ configs:
   - split: train
     path: data/train-*
 ---
+# Dataset Summary
+BCE-Arabic Layout Detection dataset contains layout annotations for Arabic documents from the BCE-Arabic corpus. It includes bounding box annotations for various region types including text regions, images, tables, charts, and graphics.
+
+## Dataset Details
+- **Size**: {size} pages
+- **Source**: BCE-Arabic Corpus
+- **Languages**: Arabic
+- **Task**: Document Layout Analysis
+
+### Supported Tasks
+- Document Layout Analysis
+- OCR Region Detection
+- Document Understanding
+
+### Contents
+Each sample contains:
+- Image: Page scan (RGB)
+- Width: Original image width
+- Height: Original image height
+- Normalized Bounding Boxes: List of normalized bounding boxes for each region
+- Region Types: List of region types for each bounding box
+- text_types: List of text types for each bounding box
+- category: Document category
+
+### Example Usage
+```python
+from datasets import load_dataset
+import json
+import cv2
+import numpy as np
+
+# Load dataset
+dataset = load_dataset("ahmedheakl/arocrbench_bce_arabic_layout")
+
+# Get a sample
+sample = dataset['train'][0]
+
+# Access image and annotations
+image = sample['image']
+width = row['width']
+height = row['height']
+bboxes = row['normalized_bounding_boxes']
+region_types = row['region_types']
+
+# Draw each bounding box
+for bbox, region_type in zip(bboxes, region_types):
+    # Convert normalized coordinates to pixel values
+    x1 = int(bbox[0] * width)
+    y1 = int(bbox[1] * height)
+    x2 = int(bbox[2] * width)
+    y2 = int(bbox[3] * height)
+    
+    color = color_map.get(region_type, (0, 0, 0))  # Default to black if unknown
+    
+    # Draw rectangle and label
+    cv2.rectangle(image, (x1, y1), (x2, y2), color, 3)
+    cv2.putText(image, region_type, (x1, y1 - 10), 
+                cv2.FONT_HERSHEY_SIMPLEX, 1, color, 2)
+cv2.imshow('image', image)
+cv2.waitKey(0)
+cv2.destroyAllWindows()
+
+```
+
+## Data Fields
+- `image`: Page image (PIL Image)
+- `width`: Image width (int)
+- `height`: Image height (int)
+- `category`: Document category (string)
+- `total_regions`: Number of regions (int)
+- `region_types`: List of region types