Animal Classification System Using Deep Learning

Overview

This project focuses on developing an Animal Classification System using deep learning and transfer learning techniques. The system is capable of identifying 15 different animal species from images with high accuracy and efficiency. It leverages the MobileNetV2 architecture for classification and features a user-friendly drag-and-drop GUI for real-time predictions.

Features

• Real-time Image Classification
• Drag-and-Drop Interface
• Confidence-Based Feedback
• High Accuracy (95.2%)
• Lightweight and Efficient Model

Technologies Used

• Deep Learning Framework: TensorFlow & Keras
• Model Architecture: MobileNetV2 (pre-trained on ImageNet)
• Programming Language: Python
• GUI Development: Tkinter & TkinterDnD
• Data Augmentation: Rotation, Zooming, Flipping

Dataset

The dataset consists of 15 classes of animals:

• Bear
• Bird
• Cat
• Cow
• Deer
• Dog
• Dolphin
• Elephant
• Giraffe
• Horse
• Kangaroo
• Lion
• Panda
• Tiger
• Zebra

Each image is resized to 224x224x3 to match MobileNetV2's input requirements.

Model Training

• Optimizer: Adam
• Learning Rate: 0.0001
• Epochs: 20
• Validation Accuracy: 95.2%
• Confidence Threshold: 70% (If below, returns "Sorry, I am unable to detect that.")

Installation

Prerequisites

Ensure you have Python installed. Recommended version: Python 3.8+

Install Dependencies

pip install tensorflow keras opencv-python pillow numpy tkinterdnd2

Clone the Repository

git clone https://github.com/shaktibiplabDev/Animal-Classification-System-Using-Deep-Learning-and-Transfer-Learning animal-classification
cd animal-classification

Run the Application

python classify.py

Usage

1. Launch the application using the command above.
2. Drag and drop an image onto the interface.
3. View the classification result and confidence score.
4. If confidence is below 70%, the system responds with: "Sorry, I am unable to detect that."

Applications

• Wildlife Monitoring: Classify animals in camera trap images.
• Education: Interactive tool for learning about animal species.
• Conservation: Helps track endangered species.

Future Improvements

• Expand dataset for better generalization.
• Improve model robustness against non-animal images.
• Support additional animal species.

References

1. Howard, A. G., et al. MobileNetV2: Inverted Residuals and Linear Bottlenecks. arXiv preprint arXiv:1801.04381 (2018).
2. Chollet, F. Deep Learning with Python. Manning Publications (2017).
3. TensorFlow Documentation: https://www.tensorflow.org/

---

Developed by Shakti Biplab