SODALMODEL 0.1.1

SODAL (Secure Object Detection and Auto-Labeling Framework): A simple yet powerful CNN wrapper for object detection, auto-labeling, training, evaluation, and model security with password protection.

🧠 SODALMODEL

A unified deep learning library for image classification, object detection, and automatic dataset labeling — powered by TensorFlow, OpenCV, ultralytics and NumPy — but easy to use with just one line of code!

---

🔥 Features

🎯 Object Detection — SVOL (Smart Vision Object Locator)

• 🚀 Pre-trained EfficientDet D0 model from TensorFlow Hub
• 🖼️ Automatic image preprocessing, bounding box detection, and drawing
• 📹 Real-time detection from webcam feed
• 🎯 High accuracy object localization on COCO dataset classes

🧠 Image Classification — SmartVisionCNN

• 🧱 Customizable convolutional neural network architecture
• 🧪 Supports training, evaluation, and accuracy/loss visualization
• 💾 Save and load trained models seamlessly
• 🔄 Easily add custom layers like Dropout for regularization

📝 Automatic Dataset Labeling — AutoLabeler

• 🤖 Automatically generate bounding box annotations for unlabeled image datasets
• 📁 Supports saving annotations in YOLO .txt and Pascal VOC .xml formats
• 🔍 Uses SVOL detection results for labeling with configurable confidence threshold
• 🎯 Requires user to provide class labels for accurate annotation generation

🔒 Model Protection — ModelProtector

• 🔐 Password-protect your trained models to restrict unauthorized access
• 🔓 Unlock models via password prompt to enable predictions and saving
• 🔒 Simple and secure file-based locking mechanism
• 🛡️ Prevents accidental or malicious model usage without permission

---

📦 Installation

Install the latest release from PyPI:

pip install SODALMODEL