🐾 Cats vs Dogs Classification using SVM

This project is part of Task 03 of the SkillCraft Technology internship program. It implements a Support Vector Machine (SVM) model to classify images of cats and dogs using pixel-level features extracted from images.

The classification model is trained on:

• Grayscale image data (resized to 64×64 pixels)
• Labeled images (filenames containing cat or dog)

The pipeline flattens images, splits them into training and validation sets, trains a Linear SVM, evaluates model performance (accuracy, confusion matrix, precision/recall), and generates predictions on the test set. Outputs include:

• Trained model (.joblib)
• Metrics report
• Submission file (.csv) in Kaggle format

📌 Technologies Used

• Python
• Pipenv
• Scikit-learn
• NumPy
• Pandas
• Pillow (PIL)
• Tqdm
• Joblib

📦 Installation

To run this project locally, follow the steps below:

1. Clone the repository:

git clone https://github.com/Agent-A345/SCT_ML_03.git

2. Install Dependencies

pip install scikit-learn numpy pandas pillow tqdm joblib

3. Run the program

python task3.py

💬 How It Works

1. Load training images from the train/ folder where filenames include class labels (cat or dog).
2. Convert each image to grayscale, resize to 64×64 pixels, and flatten into a 1D feature vector.
3. Create labeled feature vectors (X, y) for all training images.
4. Split the data into training and validation sets (80/20 stratified split).
5. Train a Linear Support Vector Machine (SVM) on the training set.
6. Evaluate the model on the validation set using accuracy, precision, recall, F1-score, and confusion matrix.
7. Load unlabeled images from the test1/ folder and generate predictions.
8. Save the trained model (model.joblib), evaluation metrics (metrics.txt), and submission file (submission.csv) to the runs/simple_svm/ directory.

📂 Dataset Used – Kaggle

📎 Dogs vs Cats Dataset
Used for training and testing the image classification model.

🙌 Acknowledgements

Thanks to SkillCraft Technology for the opportunity to work on this internship project.

License

This project is licensed under the MIT License.