The YOLO Object Detection project demonstrates a complete end-to-end workflow for detecting objects in images, videos, and real-time streams using YOLOv8 (You Only Look Once).
The project starts with pre-trained models, progresses to training a custom mini-model on the COCO dataset, and finally deploys an interactive Streamlit app for user-friendly interaction.
This project is implemented in Python and integrates YOLOv8, OpenCV, and Streamlit for building, training, evaluating, and deploying the object detection system.
- Day 1: Image detection using pre-trained YOLOv8 models
- Day 2: Object detection in uploaded videos
- Day 3: Real-time detection via webcam and multiple image inputs
- Day 4: Training a mini YOLOv8 model on a subset of the COCO dataset
- Day 5: Deployment of YOLO with a modern Streamlit web app
- Support for custom .pt weights, adjustable confidence thresholds, and output visualization
- Python 3.x
- YOLOv8 (Ultralytics) → Object detection model
- OpenCV → Image and video processing
- NumPy, Pandas → Data handling and annotations
- Matplotlib / PIL → Visualization
- Streamlit → Interactive web-based deployment
- Pyngrok → Public sharing of Streamlit apps
- Load pre-trained YOLOv8 model
- Run detection on images/videos/webcam
- Train YOLOv8 on COCO128 mini dataset
- Export trained weights (
.pt) - Deploy an interactive Streamlit app for end users
- Enhance UI with multi-page navigation in Streamlit
- Add support for live video streaming (IP cameras, RTSP feeds)
- Train on custom datasets for domain-specific detection
- Deploy on cloud platforms (AWS, GCP, Heroku) or as a Dockerized service
Abdul Rehman Ali
- 📧 Email: abdulrehman.tp.786@gmail.com