murali-09/YOLOv8n_tunned

Fine-Tuned YOLOv8n: Advanced Object Detection

This repository hosts a professionally fine-tuned YOLOv8n model optimized for high-precision object detection, specifically tailored for automated proctoring environments. The model has been exported in multiple formats to support both high-performance Python environments and cross-platform edge deployments.

🚀 Key Features

• High Accuracy: Achieved a Mean Average Precision (mAP50) of 89.5% during testing.
• Dual-Format Support: Includes PyTorch (.pt) for training/inference and ONNX for optimized deployment in web or mobile apps.
• Edge Ready: Optimized for low-latency performance on CPU, Raspberry Pi, and mobile devices.

📊 Training Results & Performance

The model was trained using the Ultralytics framework. Below is the visual breakdown of the model's performance.

Confusion Matrix

The confusion matrix below demonstrates the model's ability to distinguish between target classes with high reliability.

Training Metrics

These charts track the precision, recall, and loss metrics over the training duration, showcasing the convergence and stability of the model.

Metric	Value
Architecture	YOLOv8n (Nano)
mAP50	0.895
Model Size (PT)	~6.2 MB
Inference Speed	~10ms (on standard CPU)

🛠️ Usage

Python (Ultralytics API)

from ultralytics import YOLO

# Load the model directly from Hugging Face
model = YOLO("murali-09/YOLOv8n_tunned/model.pt")

# Perform inference on an image
results = model.predict(source="image.jpg", conf=0.25, save=True)

# View results
results[0].show()