Real-Time Object Detection using YOLOv8 & Streamlit

A Computer Vision AI application that detects objects in real time from images or live webcam feeds — powered by YOLOv8 and wrapped in a clean Streamlit web interface.

📌 Overview

Traditional object detection systems are slow, resource-heavy, and not user-friendly. This project solves that by providing a fast, accurate, and accessible web app that:

• Detects 80 common everyday objects out of the box (COCO dataset)
• Draws bounding boxes with labels and confidence scores
• Supports both image upload and live webcam modes
• Runs in the browser — no installation needed for end users

🚀 Features

• 📷 Two input modes: Image Upload (JPG/JPEG/PNG) and Live Webcam
• 🎚️ Adjustable confidence threshold slider to filter detections
• 🏷️ Real-time bounding box annotations with color-coded labels
• 📊 Detection results table showing object names and confidence scores
• 📋 Sidebar listing all 80 detectable object categories
• ⚡ Fast inference using the YOLOv8 Nano model (lightweight & efficient)
• ☁️ Fully deployable as a web app on Hugging Face Spaces

🛠️ Tech Stack

Component	Tool / Library	Purpose
Programming Language	Python 3.10+	Core development
Object Detection Model	YOLOv8 (Ultralytics)	AI model for detection
Web Framework	Streamlit	Interactive web UI
Image Processing	OpenCV (cv2)	Capture & process images/video
Image Handling	Pillow (PIL)	Open and convert image formats
Numerical Computing	NumPy	Array and matrix operations
Deployment	Hugging Face Spaces	Free cloud deployment

⚙️ How It Works

User Input (Image / Webcam)
        ↓
OpenCV reads & converts to RGB NumPy array
        ↓
YOLOv8 model runs inference (single forward pass)
        ↓
Model returns bounding boxes, class labels, confidence scores
        ↓
Annotated image + results table displayed on Streamlit UI

📦 Installation & Setup

Prerequisites

• Python 3.10 or above — download from python.org

Step 1 — Clone the Repository

git clone https://github.com/your-username/realtime-object-detection.git
cd realtime-object-detection

Step 2 — Install Dependencies

pip install -r requirements.txt

Step 3 — Run the App

streamlit run app.py

Step 4 — Open in Browser

The app opens automatically at http://localhost:8501. If it doesn't, paste the URL manually.

---

🗂️ Project Structure

realtime-object-detection/
├── app.py                  # Main Streamlit application
├── requirements.txt        # Python dependencies
└── README.md               # Project documentation

---

🧠 Detectable Objects (80 COCO Classes)

Category	Examples
People & Animals	person, cat, dog, bird, horse, cow, elephant, bear
Vehicles	car, bus, truck, motorcycle, bicycle, airplane, boat
Electronics	laptop, phone, TV, keyboard, mouse, remote
Kitchen Items	bottle, cup, fork, knife, spoon, bowl
Food	banana, apple, pizza, donut, sandwich, cake
Furniture	chair, couch, bed, dining table, toilet
Sports & Outdoors	sports ball, skateboard, surfboard, tennis racket
Other	book, clock, vase, scissors, backpack, suitcase

---

📊 Expected Results

Object	Predicted Label	Confidence
Water bottle	bottle	~90–95%
Person	person	~97%+
Mobile phone	cell phone	High
Pen (not in COCO)	may misclassify	—

Output

---

📚 References

• Ultralytics YOLOv8 Docs
• Streamlit Docs
• OpenCV Docs
• COCO Dataset
• Redmon, J. et al. (2016). You Only Look Once: Unified, Real-Time Object Detection. CVPR.
• Jocher, G. et al. (2023). Ultralytics YOLOv8. GitHub.

---