Add multi-framework dataset setup for RF-DETR, YOLOX, and YOLOv6

- Create dataset_coco/ for RF-DETR (COCO format)
- Rename dataset_split/ to dataset_yolo/ for clarity
- Add setup_datasets.py script for automated multi-format setup
- Update YOLOv6 script with correct 10-class configuration
- Update README with framework comparison and training instructions
- Update .gitignore to exclude both dataset directories

.gitignore

@@ -36,7 +36,8 @@ runs/
 # Dataset (large files)
 IMAGE/
 images/
-dataset_split/
+dataset_yolo/
+dataset_coco/
 *.jpg
 *.jpeg
 *.png

README.md

@@ -31,7 +31,9 @@ This repository contains a complete wood defect detection system using YOLOX/YOL
 - Valid: 2,027 images
 - Test: 2,029 images
 
-**Format**: YOLO format (images/ and labels/ subdirectories with data.yaml configuration)
+**Formats Available**:
+- `dataset_coco/` → COCO format for RF-DETR
+- `dataset_yolo/` → YOLO format for YOLOX, YOLOv6, YOLOv8
 
 ## 🚀 Quick Start
 
@@ -48,21 +50,19 @@ source .venv/bin/activate
 
 # Install dependencies
 pip install -U pip
-pip install ultralytics gradio
+pip install ultralytics gradio rfdetr
 ```
 
-### 2. Download Dataset
+### 2. Setup Datasets
-
-The dataset is not included in the repository due to size. Download from Kaggle and organize as follows:
 
 ```bash
-# Download from Kaggle (requires Kaggle API)
+# Download dataset from Kaggle (requires Kaggle API)
 kaggle datasets download -d kirs0816/wood-surface-defects
 unzip wood-surface-defects.zip
 
-# Run the dataset preparation script
+# Create multi-format datasets
-python split_coco_dataset.py
+python split_coco_dataset.py  # Creates dataset_yolo/
-python reorganize_dataset.py
+python setup_datasets.py      # Creates dataset_coco/ and updates configs
 ```
 
 ### 3. Launch Annotation GUI
@@ -77,10 +77,38 @@ Open http://localhost:7860 in your browser to access the web-based annotation in
 - Manual annotation tools
 - Real-time result visualization
 
-### 4. Train Model
+### 4. Train Models
 
+Choose from three different frameworks:
+
+#### RF-DETR (Highest accuracy, slower training)
 ```bash
-python train_yolox.py --dataset-dir dataset_split --model yolox-nano --epochs 5 --batch-size 4
+python train_rfdetr.py \
+  --dataset-dir dataset_coco \
+  --output-dir runs/rfdetr_medium \
+  --model medium \
+  --epochs 50 \
+  --batch-size 4 \
+  --grad-accum-steps 4 \
+  --lr 1e-4
+```
+
+#### YOLOX (Balanced performance/speed)
+```bash
+python train_yolox.py \
+  --dataset-dir dataset_yolo \
+  --model yolox-nano \
+  --epochs 50 \
+  --batch-size 8
+```
+
+#### YOLOv6 (Fastest, edge-optimized)
+```bash
+python train_yolov6.py \
+  --dataset-dir dataset_yolo \
+  --model yolov6n \
+  --epochs 50 \
+  --batch-size 8
 ```
 
 ## 📁 Project Structure
@@ -88,11 +116,23 @@ python train_yolox.py --dataset-dir dataset_split --model yolox-nano --epochs 5
 ```
 saw_mill_knot_detection/
 ├── annotation_gui.py          # Gradio web interface for annotation
+├── train_rfdetr.py           # RF-DETR training script
 ├── train_yolox.py            # YOLOX training script
+├── train_yolov6.py           # YOLOv6 training script
+├── setup_datasets.py         # Multi-format dataset setup script
 ├── split_coco_dataset.py     # Dataset splitting utility
-├── reorganize_dataset.py      # Dataset reorganization to YOLO format
 ├── config.py                 # Configuration settings
-├── dataset_split/             # Training data (excluded from git)
+├── dataset_coco/             # RF-DETR dataset (COCO format)
+│   ├── train/
+│   │   ├── *.jpg             # Training images
+│   │   └── _annotations.coco.json
+│   ├── valid/
+│   │   ├── *.jpg             # Validation images
+│   │   └── _annotations.coco.json
+│   └── test/
+│       ├── *.jpg             # Test images
+│       └── _annotations.coco.json
+├── dataset_yolo/             # YOLOX/YOLOv6/YOLOv8 dataset (YOLO format)
 │   ├── train/
 │   │   ├── images/           # Training images
 │   │   └── labels/           # YOLO format labels
@@ -104,17 +144,20 @@ saw_mill_knot_detection/
 │   │   └── labels/           # YOLO format labels
 │   └── data.yaml             # YOLO dataset configuration
 ├── runs/                     # Training outputs (excluded from git)
-│   └── yolox_training/
-│       └── training/
-│           └── weights/
-│               ├── best.pt   # Best model weights
-│               └── last.pt   # Latest model weights
 ├── bbox_coco_dataset.json     # Original COCO annotations
 ├── requirements.txt           # Python dependencies
 ├── .gitignore                # Excludes large data files
 └── README.md                 # This file
 ```
 
+## 🤖 Framework Comparison
+
+| Framework | Accuracy | Speed | Memory | Deployment | Best For |
+|-----------|----------|-------|--------|------------|----------|
+| **RF-DETR** | ⭐⭐⭐⭐⭐ | ⭐⭐ | ⭐⭐⭐ | CPU/GPU | Highest accuracy, research |
+| **YOLOX** | ⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐ | Edge devices | Balanced performance |
+| **YOLOv6** | ⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐ | Mobile/Edge | Fast inference, production |
+
 ## 🛠️ Usage Guide
 
 ### Annotation GUI Features

setup_datasets.py

@@ -0,0 +1,111 @@
+#!/usr/bin/env python3
+"""
+Setup multi-format datasets for different model frameworks.
+
+Creates:
+- dataset_coco/ for RF-DETR (COCO format)
+- dataset_yolo/ for YOLOX/YOLOv6/YOLOv8 (YOLO format)
+
+Usage:
+    python setup_datasets.py
+"""
+
+import json
+import shutil
+from pathlib import Path
+
+
+def setup_coco_dataset():
+    """Set up COCO format dataset for RF-DETR."""
+    print("Setting up COCO format dataset...")
+
+    coco_dir = Path("dataset_coco")
+    yolo_dir = Path("dataset_yolo")
+
+    if not yolo_dir.exists():
+        print("Error: dataset_yolo/ not found. Run split_coco_dataset.py first!")
+        return False
+
+    # Create COCO directories
+    for split in ["train", "valid", "test"]:
+        split_dir = coco_dir / split
+        split_dir.mkdir(parents=True, exist_ok=True)
+
+        # Copy images from YOLO dataset
+        yolo_images = yolo_dir / split / "images"
+        if yolo_images.exists():
+            for img_file in yolo_images.glob("*"):
+                shutil.copy2(img_file, split_dir)
+
+        # Copy COCO annotations
+        coco_ann = yolo_dir / split / "_annotations.coco.json"
+        if coco_ann.exists():
+            shutil.copy2(coco_ann, split_dir)
+
+    print(f"COCO dataset created at: {coco_dir}")
+    return True
+
+
+def update_yolov6_data_config():
+    """Update YOLOv6 data config to use correct number of classes."""
+    print("Updating YOLOv6 data configuration...")
+
+    # Load the COCO annotations to get class information
+    coco_file = Path("dataset_yolo/train/_annotations.coco.json")
+    if not coco_file.exists():
+        print("Warning: Cannot find COCO annotations to update YOLOv6 config")
+        return
+
+    with coco_file.open('r') as f:
+        data = json.load(f)
+
+    categories = data['categories']
+    nc = len(categories)
+    names = [cat['name'] for cat in categories]
+
+    # Update the YOLOv6 training script
+    yolov6_script = Path("train_yolov6.py")
+    if yolov6_script.exists():
+        content = yolov6_script.read_text()
+
+        # Replace hardcoded nc: 1 and names: ['knot']
+        old_config = "nc: 1\nnames: ['knot']"
+        new_config = f"nc: {nc}\nnames: {names}"
+
+        if old_config in content:
+            content = content.replace(old_config, new_config)
+            yolov6_script.write_text(content)
+            print(f"Updated YOLOv6 script with {nc} classes: {names}")
+        else:
+            print("YOLOv6 config already updated or not found")
+
+
+def main():
+    print("Setting up multi-format datasets for different ML frameworks...\n")
+
+    # Setup COCO format for RF-DETR
+    if setup_coco_dataset():
+        print("✅ COCO format dataset ready for RF-DETR")
+    else:
+        print("❌ Failed to setup COCO dataset")
+        return
+
+    # Update YOLOv6 configuration
+    update_yolov6_data_config()
+
+    print("\n" + "="*60)
+    print("DATASET SETUP COMPLETE!")
+    print("="*60)
+    print("Available datasets:")
+    print("  📁 dataset_coco/  → RF-DETR (COCO format)")
+    print("  📁 dataset_yolo/  → YOLOX, YOLOv6, YOLOv8 (YOLO format)")
+    print()
+    print("Training commands:")
+    print("  🔶 RF-DETR:  python train_rfdetr.py --dataset-dir dataset_coco --output-dir runs/rfdetr")
+    print("  🔵 YOLOX:    python train_yolox.py --dataset-dir dataset_yolo --model yolox-nano")
+    print("  🟡 YOLOv6:   python train_yolov6.py --dataset-dir dataset_yolo --model yolov6n")
+    print("="*60)
+
+
+if __name__ == "__main__":
+    main()

train_yolov6.py

@@ -68,8 +68,8 @@ def train_yolov6(
         f.write(f"""train: {train_dir.absolute()}
 val: {valid_dir.absolute()}
 
-nc: 1
+nc: 10
-names: ['knot']
+names: ['Live knot', 'Dead knot', 'Knot with crack', 'Crack', 'Resin', 'Marrow', 'Quartzity', 'Knot missing', 'Blue stain', 'Overgrown']
 """)
     
     print(f"\n{'='*60}")