README.md

---
license: mit
library_name: keras
tags:
- tensorflow
- keras
- cnn
- computer-vision
- image-classification
- dogs-vs-cats
- binary-classification
---

# 🐶🐱 Dog vs Cat Image Classifier

## 📌 Overview

This repository contains a Convolutional Neural Network (CNN) developed using TensorFlow/Keras for binary image classification. The model classifies input images as either **Dog** or **Cat**.

This project was created as part of my deep learning portfolio to demonstrate CNN design, model training, evaluation, and deployment practices.

---

## 🧠 Model Details

- **Framework:** TensorFlow / Keras
- **Architecture:** Convolutional Neural Network (CNN)
- **Task:** Binary Image Classification
- **Classes:** Dog, Cat
- **Input Size:** 256 × 256 × 3
- **Epochs:** 10
- **Validation Accuracy:** ~95–96%

---

## 📊 Training

The model was trained on the Kaggle Dogs vs Cats dataset.

Training included:

- Image preprocessing
- CNN feature extraction
- Binary classification using a sigmoid output layer
- Model evaluation using accuracy and loss metrics

---

## 🚀 Usage

```python
from tensorflow.keras.models import load_model

model = load_model("cnn_model.keras")
```

---

## 📈 Results

- Validation Accuracy: **95–96%**
- Binary Classification
- TensorFlow/Keras Implementation

---

## 📚 Future Improvements

- Transfer Learning (EfficientNet / ResNet50)
- Data Augmentation
- Hyperparameter Tuning
- Grad-CAM Visualization
- Streamlit Deployment

---

## 👨‍💻 Author

**Vertika**

GitHub:
https://github.com/vertika13122007-tech