-# 🐾 Animal Vision System: Classification & Identification

Dairy Cattle Breed Classifier

Dog Facial Similarity Identification

An advanced computer vision project focused on animal identification and breed classification. This repository features two primary systems: a Dairy Cattle Breed Classifier and a Dog Facial Similarity Identification engine.

---

🌟 Key Features

🐄 Cattle Breed Classification

• Transfer Learning: Utilizes pre-trained MobileNetV2 for high-accuracy feature extraction with minimal computational overhead.
• Breed Detection: Specifically optimized for identifying dairy cattle breeds like Gir and Sahiwal.
• Automated Pipeline: Includes end-to-end workflows from image scraping to model deployment.

🐕 Dog Facial Identification

• Face Similarity Matching: Calculates similarity scores between dog faces to determine if they belong to the same individual.
• Stanford Dogs Integration: Uses the comprehensive Stanford Dogs dataset for robust model training and testing.
• Interactive UI: Features an IPyWidgets-based interface within Jupyter Notebooks for real-time comparison and visualization.

🛠️ Data Infrastructure

• Multi-Source Scraping: Integrated icrawler workflows to download images from Bing, Google, and Baidu.
• Intelligent Splitting: Automated scripts to organize raw data into structured Train, Validation, and Test directories.
• Augmentation: Implements real-time data augmentation (flips, rotations, zooms) to enhance model generalization.

---

🏗️ Project Structure

Animal-Vision-System/
├── cattle_model.ipynb         # Cattle classification training & evaluation
├── dog_face_detection_V1.ipynb # Dog facial identification & interactive matching
├── datasets/                  # (Generated) Local storage for training data
└── README.md                  # Project documentation

---

🚀 Getting Started

Prerequisites

• Python 3.8 or higher
• GPU recommended for training (e.g., NVIDIA T4 in Google Colab)

Installation

1. Clone the repository:

   git clone https://github.com/Rak2k6/Animal-Vision-System---Classification-Identification.git
   cd Animal-Vision-System---Classification-Identification

2. Install required dependencies:

   pip install tensorflow icrawler scikit-learn matplotlib pillow ipywidgets

---

📊 Methodology

1. Data Acquisition

The system leverages the icrawler library to build custom datasets. For example, the cattle model uses search keywords like "Murrah" and "Gir Cattle" to pull diverse samples across global search engines.

2. Model Architecture

• Base: MobileNetV2 (Frozen weights from ImageNet).
• Top: Global Average Pooling -> Dropout (0.3) -> Dense (Sigmoid/Softmax).
• Optimizer: Adam.
• Loss: Binary/Categorical Crossentropy.

3. Training & Validation

Models are trained with Early Stopping to prevent overfitting, monitoring validation loss to restore the best performing weights.

---

🖥️ Usage

Cattle Classification

Open cattle_model.ipynb and run the cells sequentially to:

1. Scrape images for desired breeds.
2. Split data into directories.
3. Train the MobileNetV2 model.
4. Save the model as .h5 for deployment.

Dog Face Identification

Open dog_face_detection_V1.ipynb to:

1. Download and extract the Stanford Dogs dataset.
2. Run the interactive comparison widget.
3. Upload two dog images to see their similarity score and identification decision.

---

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

Developed by Rak2k6