Bird Species Classifier

The Bird Species Classifier is an application built using a Convolutional Neural Network (CNN) to classify images of birds into one of 525 different species. It allows users to upload an image of a bird and receive a prediction of the bird species along with the confidence score.

Key Features:

• Image Upload: Users can upload an image of a bird directly to the application.

• Classification: The application uses a trained CNN model to classify the uploaded image into one of 525 bird species.

• Confidence Score: Along with the predicted bird species, the application provides a confidence score indicating the certainty of the prediction.

Model Training:

The CNN model used in this application has been trained on a dataset containing images of 525 different bird species. The model has been trained using various optimizers, including Adam, Nadam, SGD, RMSprop, and Adagrad, to optimize its performance.

Custom Metrics:

The model's performance is evaluated using precision, recall, and F1-score metrics, which provide measures of the model's accuracy in classifying bird species.

Usage:

1. Upload Image: Click on the "Choose an image..." button to upload an image of a bird.

2. Prediction: After uploading the image, click on the "Predict" button to classify the bird species.

3. Result: The application will display the predicted bird species along with the confidence score.

Requirements:

• Python 3.x
• Streamlit
• TensorFlow
• PIL (Python Imaging Library)
• NumPy
• scikit-learn

Running the Application:

1. Install the required dependencies mentioned above.

2. Clone the repository or download the provided files.

3. Run the Streamlit app script (app.py or any appropriate filename) using the following command:

   streamlit run app.py

4. Access the application through the provided URL in the terminal.

Dataset: https://www.kaggle.com/datasets/gpiosenka/100-bird-species/code

The dataset used to train the model contains a diverse collection of bird images spanning 525 different species. Each image is labeled with its corresponding bird species to facilitate supervised learning.

Model Training:

The model is trained using the following steps:

• Data Augmentation: Images are augmented using techniques such as rotation, shifting, shearing, zooming, and flipping to increase the diversity of the training data.

• Model Architecture: The CNN model consists of multiple convolutional layers followed by max-pooling layers to extract features from the input images.

• Optimizer: The model is compiled using various optimizers, including Adam, Nadam, SGD, RMSprop, and Adagrad, to optimize its performance.

Evaluation:

The trained model is evaluated on a separate test dataset to assess its performance. Metrics such as accuracy, precision, recall, and F1-score are computed to measure the model's effectiveness in classifying bird species.

Future Enhancements:

1. Improving Model Accuracy: Continuously train and fine-tune the model to improve its accuracy in classifying bird species.

2. Interactive Visualization: Enhance the user interface with interactive features such as displaying similar bird species, bird habitat information, and bird calls.

3. Multi-Image Classification: Extend the application to support batch processing of multiple bird images for classification.

4. Mobile Application: Develop a mobile version of the application for users to classify bird species on the go.

Acknowledgments:

• The dataset used in this project is sourced from Kaggle and https://www.kaggle.com/datasets/gpiosenka/100-bird-species/code.

• Special thanks to Ayan Sar and Purvika Joshi for their contributions to the project.

Contributors:

• Subhangi Sati

• Purvika Joshi

• Ayan Sar