dl_xview_yolo

🛰️ DL_XVIEW • Object Detection in Satellite Imagery

Issues

Deep Learning–based Object Detection system for satellite imagery using YOLOv8.
This project applies state-of-the-art computer vision models to detect objects such as airplanes, ships, vehicles, bridges, and more from aerial or satellite images.
Data sources include xView and DOTA datasets.

Goal: Detect and classify multiple objects in high-resolution satellite imagery
Model: YOLOv8 (with Oriented Bounding Box / OBB support)

Features

Object detection using YOLOv8 (Ultralytics)
Dataset conversion utilities for xView / DOTA → YOLO format (convert_all_to_yolo.py)
Interactive web-based prediction UI (ui/index.html)
Utility scripts: train_yolo.py, predict_yolo.py, summarize_run.py
Easy path and parameter customization

🗂️ Project Structure

dl_xview_yolo/
├── scripts/
│   ├── convert_all_to_yolo.py     # Converts datasets to YOLO format
│   ├── train_yolo.py              # Training script
│   ├── predict_yolo.py            # Inference script
│   ├── summarize_run.py           # Summarizes training metrics
│   └── prepare_dota_v1_to_yolo_obb.py
│
├── ui/
│   └── index.html                 # Interactive front-end for predictions
│
├── YOLOv8/                        # YOLOv8 models / configs
├── .gitignore
├── LICENSE
├── README.md
└── requirements.txt


## ⚙️ Installation & Setup

### 1️⃣ Environment Preparation

> Recommended: Python 3.10 or higher

```bash
git clone https://github.com/Yigtwxx/dl_xview_yolo.git
cd dl_xview_yolo
python -m venv venv
venv\Scripts\activate   # Windows
pip install -r requirements.txt

If you don’t have a requirements.txt file, install manually:

pip install ultralytics opencv-python pillow tqdm numpy torch torchvision torchaudio matplotlib

Dataset Preparation

1️⃣ Supported Datasets

2️⃣ Convert to YOLO Format

python scripts/convert_all_to_yolo.py --src "data/raw" --out "data/yolo_data" --copy

Explanation:

Converts raw .txt annotations into YOLO-style labels
Saves output to data/yolo_data
Copies images when --copy flag is used

Training

python scripts/train_yolo.py --data "configs/data.yaml" --model "yolov8n.pt" --epochs 40 --imgsz 1024

Model checkpoints are saved under runs/train/...
You can monitor metrics via results.png (mAP, loss curves, etc.)

Inference

Command Line

python scripts/predict_yolo.py --weights "runs/train/xview-yolo/weights/best.pt" --source "data/test_images"

Web UI

# If integrated with FastAPI / Streamlit UI:
python scripts/predict_yolo.py --ui

Then open your browser at 👉 http://127.0.0.1:7860/

🎨 User Interface (UI)

Open ui/index.html directly in your browser to visualize predictions.

💫 Modern design: world-themed background & glass-panel overlay 📂 Upload: drag-and-drop or select image 🧠 Backend: connects directly to YOLOv8 model for real-time inference

📊 Example Results

Metric	Value
mAP@0.5	0.54
mAP@0.5-0.95	0.36
Precision	0.67
Recall	0.71

Metrics may vary depending on dataset and training configuration. See runs/train/.../results.png for detailed training graphs.

📦 Model Weights

If your trained model (best.pt) is smaller than 100 MB, you can store it in:

runs/train/xview-yolo/weights/best.pt

Otherwise, upload to Google Drive and link it here:

[📁 Download best.pt](https://drive.google.com/your-model-link)

🧩 Requirements

🧠 Python ≥ 3.10
⚡ PyTorch ≥ 2.0
🎯 Ultralytics YOLOv8
📸 OpenCV, Pillow
📊 Matplotlib

⚠️ Notes

Training datasets (data/, runs/, venv/) are excluded from Git using .gitignore.
To resume training from the last checkpoint:

python scripts/train_yolo.py --resume

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.

💬 Author

Yiğit Erdoğan

📧 Mail: yigiterdogan6@icloud.com

🧠 Focus Areas: Deep Learning • Computer Vision • Data Science

Contact

LinkedIn: www.linkedin.com/in/yiğit-erdoğan-ba7a64294

⭐ Support

If you find this project useful, please consider giving it a ⭐ Star on GitHub!

🌐 Live Project Page: https://yigtwxx.github.io/dl_xview_yolo

Contact

LinkedIn: www.linkedin.com/in/yiğit-erdoğan-ba7a64294

git clone https://github.com/Yigtwxx/dl_xview_yolo.git

This site is open source. Improve this page.