Update README.md

README.md

@@ -12,7 +12,7 @@ tags:
 - synthetic-data
 pretty_name: Synthetic Bank Statement Table Detection Dataset
 size_categories:
-- n<10k
+- 10K<n<100K
 dataset_info:
   features:
   - name: image
@@ -50,10 +50,10 @@ A synthetically generated collection of bank statement images with pixel-perfect
 | | |
 |---|---|
 | **Task** | Object Detection → Table Detection |
-| **Format** | PNG images + YOLO-format `.txt` labels |
+| **Format** | Parquet shards with embedded PNG images + inline bounding-box annotations (Hugging Face `datasets` format) |
 | **Classes** | 1 (`0` = Table) |
-| **Total images** | 8475 |
-| **Splits** | None yet — single unsplit set |
+| **Total images** | 15,187 |
+| **Splits** | None yet — single unsplit `train` set |
 | **Source** | 100% synthetic, generated with Python + ReportLab |
 | **Real data?** | ❌ No real bank statements, customers, or financial records |
 | **License** | MIT |
@@ -74,8 +74,8 @@ Because every annotation is produced during rendering, the resulting labels are
 
 Each sample consists of:
 
-- A synthetic bank statement image
-- One YOLO annotation file
+- A synthetic bank statement image (embedded directly in the parquet row)
+- An inline `objects` field holding the bounding box(es) and class(es)
 - A bounding box surrounding the complete transaction table
 
 ### Layout diversity
@@ -95,37 +95,30 @@ The generator produces a wide variety of layouts, including:
 ## Folder Structure
 
 ```text
-table_detection/
-├── images/
-│   ├── anon_stmt_001_p1.png
-│   ├── anon_stmt_001_p2.png
-│   ├── anon_stmt_002_p1.png
-│   └── ...
-└── labels/
-    ├── anon_stmt_001_p1.txt
-    ├── anon_stmt_001_p2.txt
-    ├── anon_stmt_002_p1.txt
-    └── ...
+bank-statement-detection/
+└── data/
+    ├── train-00000-of-00008.parquet
+    ├── train-00001-of-00008.parquet
+    ├── ...
+    └── train-00007-of-00008.parquet
 ```
 
-Each image has a matching YOLO label file with the same base filename.
+There are no loose image/label files in this repo — every sample (image + its annotation) lives inline in these parquet shards, loadable directly with the `datasets` library (see below). Each row has two fields:
 
-**Filename convention:** `anon_stmt_<statement_id>_p<page_number>`
+- `image` — the rendered statement page (decoded to a PIL image)
+- `objects` — a struct with `bbox` (list of `[x_center, y_center, width, height]`, normalized) and `category` (list of class ids)
 
-- `anon_` — synthetic/anonymized document
-- `stmt_<statement_id>` — unique statement identifier
-- `p<page_number>` — page number
-
-Multi-page statements are stored as separate image/label pairs.
+Multi-page statements are stored as separate rows, one per page.
 
 ---
 
 ## Label Format
 
-Standard YOLO object detection format:
+Each row's `objects` field uses standard YOLO-style coordinates, stored inline rather than as a separate `.txt` file:
 
 ```text
-class_id center_x center_y width height
+objects.category[i] = class_id
+objects.bbox[i]     = [center_x, center_y, width, height]
 ```
 
 - `class_id`: `0` (Table)
@@ -139,47 +132,19 @@ class_id center_x center_y width height
 
 ## Data Example
 
-The examples below show raw statement images and the contents of their matching YOLO label files.
-
-<table>
-<tr>
-<th>Image</th>
-<th>Image</th>
-</tr>
-<tr>
-<td>
-
-![](images/anon_stmt_001_p1.png)
-
-</td>
-<td>
-
-![](images/anon_stmt_002_p1.png)
+Loading a single row shows the image and its inline annotation together:
 
-</td>
-</tr>
-<tr>
-<th>Label</th>
-<th>Label</th>
-</tr>
-<tr>
-<td>
-
-```text
-0 0.503 0.548 0.856 0.701
-```
+```python
+from datasets import load_dataset
 
-</td>
-<td>
+ds = load_dataset("Panhapich/bank-statement-detection", split="train")
+sample = ds[0]
 
-```text
-0 0.498 0.541 0.842 0.688
+sample["image"]        # PIL.Image, the rendered statement page
+sample["objects"]
+# {'bbox': [[0.503, 0.548, 0.856, 0.701]], 'category': [0]}
 ```
 
-</td>
-</tr>
-</table>
-
 Each image typically contains one bounding box representing the entire transaction table.
 
 ---
@@ -188,14 +153,10 @@ Each image typically contains one bounding box representing the entire transacti
 <summary><strong>Load with the Hugging Face Hub</strong></summary>
 
 ```python
-from huggingface_hub import snapshot_download
+from datasets import load_dataset
 
-local_dir = snapshot_download(
-    repo_id="Panhapich/Panhapich/bank-statement-detection",
-    repo_type="dataset"
-)
-
-print(local_dir)
+ds = load_dataset("Panhapich/bank-statement-detection", split="train")
+print(ds)
 ```
 
 </details>
@@ -203,6 +164,29 @@ print(local_dir)
 <details>
 <summary><strong>Train with Ultralytics YOLO</strong></summary>
 
+YOLO expects images and `.txt` labels as loose files, so export the parquet rows to disk first:
+
+```python
+from pathlib import Path
+from datasets import load_dataset
+
+ds = load_dataset("Panhapich/bank-statement-detection", split="train")
+
+img_dir, lbl_dir = Path("table_detection/images"), Path("table_detection/labels")
+img_dir.mkdir(parents=True, exist_ok=True)
+lbl_dir.mkdir(parents=True, exist_ok=True)
+
+for i, sample in enumerate(ds):
+    sample["image"].save(img_dir / f"{i:06d}.png")
+    lines = [
+        f"{cls} {' '.join(map(str, bbox))}"
+        for bbox, cls in zip(sample["objects"]["bbox"], sample["objects"]["category"])
+    ]
+    (lbl_dir / f"{i:06d}.txt").write_text("\n".join(lines))
+```
+
+Then train as usual:
+
 ```yaml
 path: ./table_detection
 
@@ -238,7 +222,7 @@ Every annotation is generated automatically during document creation:
 1. A synthetic bank statement is procedurally generated.
 2. ReportLab renders the transaction table.
 3. The exact table coordinates are captured during rendering.
-4. The coordinates are converted into normalized YOLO format and written to the matching label file.
+4. The coordinates are converted into normalized YOLO format and stored inline as the row's `objects` field.
 
 Because annotations originate directly from the rendering engine, they provide pixel-perfect ground truth with zero manual labeling.