Cat and Dog Image Classifier using Convolutional Neural Network (CNN)

https://classify-cat-dog.streamlit.app/

This project implements a Convolutional Neural Network (CNN) model to classify images as either cats or dogs. The model is trained on a dataset containing labeled images of cats and dogs, and achieves competitive performance in distinguishing between the two classes.

Overview

This CNN model is built using [insert framework/library here, e.g., TensorFlow, PyTorch] and trained on a dataset comprising thousands of cat and dog images. The goal of the model is to accurately predict whether a given image contains a cat or a dog.

Model Architecture

The CNN architecture used for this project consists of several convolutional layers followed by max-pooling layers to extract and learn features from the input images. The final layers are fully connected to make predictions based on the learned features.

• Input Layer: Accepts input images of size [specify dimensions].
• Convolutional Layers: [Brief description of layers and filters used].
• Pooling Layers: Max-pooling layers to downsample feature maps.
• Fully Connected Layers: Dense layers for classification.

Dataset

The model is trained on a dataset sourced from Kaggle. It includes a balanced collection of cat and dog images, each labeled accordingly.

Model Definition

• Activation Function: ReLU is used in the convolutional layers.
• Output Activation Function: Sigmoid is used in the final dense layer for binary classification.

Training Configuration

• Optimizer: Adam optimizer is used ('adam').
• Loss Function: Binary Cross-Entropy ('binary_crossentropy') is used.
• Metrics: Accuracy is used to evaluate the model's performance.
• Training Steps: 11 epochs with a batch size of 64.

Future Improvements

Potential enhancements and future work include:

• Experimenting with different architectures (e.g., transfer learning).
• Enhancing dataset diversity.
• Improving model robustness and efficiency.

Screenshot

Model