---

🧠 YOLOv8 Object Detection Project

📌 Project Overview

This project demonstrates how to train and use a YOLOv8 (You Only Look Once v8) model for custom object detection using the Ultralytics YOLO library.

The model is trained on a custom dataset and can perform real-time object detection on:

• Images 🖼️
• Videos 🎥
• Webcam streams 📷

YOLOv8 is a state-of-the-art, one-stage object detection architecture that provides high accuracy and real-time performance for edge and embedded devices.

---

🧱 Folder Structure

Ensure your dataset is organized as follows before training:

If you’re not familiar with how to create your own dataset — including image labeling and annotation generation — you can refer to the following guide:

dataset/
├── images/
│   ├── train/
│   └── val/
├── labels/
│   ├── train/
│   └── val/
└── data.yaml

💻 Choose a Code Editor

You can use any IDE or code editor such as:

• Visual Studio Code (VS Code)
• Eclipse
• PyCharm

---

🧩 Setting Up Your Environment

1️⃣ Create and Activate a Virtual Environment (Recommended)

# Create virtual environment
python -m venv venv

For Windows:

venv\Scripts\activate

For Linux/Mac:

source venv/bin/activate

---

2️⃣ Install Dependencies

Make sure your virtual environment is active, then install the required libraries:

pip install ultralytics torch torchvision opencv-python pyyaml

Alternatively, create a requirements.txt file and run:

pip install -r requirements.txt

---

🚀 Training the Model

You can train your YOLOv8 model using the modeltraining.py script or directly from the terminal.

Option 1: Using Python Script

from ultralytics import YOLO

# Load a pre-trained YOLOv8 model (e.g., small version)
model = YOLO("yolov8s.pt")

# Train the model on your custom dataset
model.train(data="dataset/data.yaml", epochs=100, imgsz=640, batch=16, name="custom_yolo_train")

Option 2: Using Command Line

yolo detect train data=dataset/data.yaml model=yolov8s.pt epochs=100 imgsz=640 batch=16 name=custom_yolo_train

---

🏋️‍♀️ Training Output

For this project, I trained the YOLOv8 model to detect three custom object classes. Below are the results obtained after training, and I’ve also included my training outputs and evaluation results under the runs/ directory.

After training, YOLOv8 will automatically create a directory:

runs/train/custom_yolo_train/

Inside it, you’ll find:

• weights/best.pt → Best performing model
• weights/last.pt → Model state from last epoch
• results.csv → Training logs
• results.png → Training graphs (loss, mAP, precision, recall)

---

🧠 Running Inference

Once you have your trained weights (best.pt), use the following script to detect objects on images, videos, or webcam:

from ultralytics import YOLO

# Load trained model
model = YOLO("runs/train/custom_yolo_train/weights/best.pt")

# Run inference on an image
results = model.predict(source="test_images/image1.jpg", conf=0.5, show=True)

Other Options:

• Images: source="path/to/image.jpg"
• Videos: source="path/to/video.mp4"
• Webcam: source=0

Detected results will be saved in:

runs/detect/predict/

---

📊 Results & Evaluation

After training, YOLOv8 automatically computes performance metrics such as:

• mAP (Mean Average Precision)
• Precision
• Recall
• F1 Score

You can visualize these metrics and training curves inside:

runs/train/custom_yolo_train/

---

📁 Model Weights

Best weights will be saved at:

runs/train/custom_yolo_train/weights/best.pt

Use this file for inference, evaluation, or deployment.

---

🧾 Optional: Export the Model

YOLOv8 supports exporting to multiple formats for deployment:

yolo export model=runs/train/custom_yolo_train/weights/best.pt format=onnx

Available formats: onnx, torchscript, coreml, tflite, openvino, engine (TensorRT).

---

🛠️ Troubleshooting

Problem	Possible Cause	Solution
No labels found	Label files missing or not matching image names	Ensure each image has a .txt label with the same name
CUDA not available	PyTorch not installed with GPU support	Reinstall PyTorch with CUDA enabled
Low mAP / accuracy	Poor labeling or small dataset	Increase dataset size or improve label quality
Out of memory (OOM)	Batch size too large for GPU	Reduce batch size or use smaller image size

---

📚 References

• Ultralytics YOLOv8 Documentation
• YOLO Paper Series (v1–v8)
• PyTorch Official Docs

---

SAI CHAITANYA KANCHARANA 📧kancharanasaichaitanya@gmail.com 🌐https://github.com/saichaitanya-c/

---