CoolFace
Apppublic

isamadafridi/obstacle_detection

sourceHugging Faceupdated 4mo agoView on Hugging Face
0likes
App README

YOLOv8 Obstacle Detection System

A real-time obstacle detection system built using YOLOv8 and trained on a custom obstacle detection dataset. This project detects multiple real-world objects such as vehicles, pedestrians, road obstacles, traffic signs, poles, stairs, dustbins, and more.

The system is deployed on Hugging Face Spaces using Gradio for interactive inference.

🚀 Live Demo

👉 Hugging Face Demo

📂 Dataset

Dataset used for training:

👉 Obstacle Detection Dataset (Kaggle)

Dataset Statistics

  • —25 object classes
  • —Real-world outdoor obstacle images
  • —Annotated bounding boxes
  • —YOLO format annotations
  • —Suitable for autonomous navigation systems

🧠 Model Architecture

This project uses YOLOv8 for object detection.

Model Variant

  • —YOLOv8n (Nano version)

Why YOLOv8n?

YOLOv8n was selected because it provides:

  • —Fast inference speed
  • —Lightweight architecture
  • —Low memory usage
  • —Real-time detection capability
  • —Easy deployment

⚙️ Training Configuration

ParameterValue
ModelYOLOv8n
Epochs50
Image Size512 × 512
Batch Size8
OptimizerMuSGD
FrameworkPyTorch
GPUTesla T4

🔥 Data Augmentation

The model was trained using YOLOv8 built-in augmentations:

  • —Mosaic augmentation
  • —Horizontal flip
  • —HSV color augmentation
  • —Translation augmentation
  • —Scaling augmentation
  • —Blur augmentation
  • —CLAHE enhancement

These augmentations improve model robustness and generalization.

📈 Model Performance

MetricScore
mAP500.902
mAP50-950.728
Precision0.884
Recall0.834

✨ Features

  • —Real-time obstacle detection
  • —Multi-class object detection
  • —Interactive web interface
  • —YOLOv8 inference
  • —Upload and analyze images
  • —Lightweight deployment
  • —Hugging Face Spaces integration

🌐 Tech Stack

Frontend

  • —Gradio

Backend

  • —Python 3.13

Deep Learning

  • —PyTorch
  • —Ultralytics YOLOv8

Deployment

  • —Hugging Face Spaces

📁 Project Structure

bash
project/
│
├── app.py
├── best.pt
├── requirements.txt
└── README.md

📦 Installation

Clone Huggingface Repository

bash
git clone https://huggingface.co/spaces/isamadafridi/obstacle_detection
cd your-repository-name

Github Repository

🔗.Github Repo

📥 Install Dependencies

bash
pip install -r requirements.txt

▶️ Run Application

bash
python app.py

🧪 Sample Inference

python
from ultralytics import YOLO

model = YOLO("best.pt")

results = model.predict(
    source="image.jpg",
    conf=0.25,
    save=True
)

🌍 Deployment

This project is deployed on Hugging Face Spaces.

👉 Live Application

🚀 Future Improvements

  • —Video stream detection
  • —Webcam integration
  • —TensorRT optimization
  • —Mobile deployment
  • —Edge AI optimization
  • —Real-time surveillance system

👨‍💻 Developer

Abdul Samad Afridi

📧 Email: isamadafridi@gmail.com

📜 License

This project is intended for educational and research purposes.

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference