Expressions Decoded: Image Classification Modeling

Priya Iragvarupu, Francesca Scipioni, Sowjanya Yaddanapudi, Omar Zu'bi

About The Project

Convolutional Neural Networks (CNNs) are a class of deep learning models particularly suited for processing structured grid-like data, such as images and videos. They are designed to automatically and adaptively learn spatial hierarchies of features through a series of convolutional layers. Each layer consists of filters that convolve over input data, capturing local patterns and features. CNNs are widely used in tasks such as image classification, object detection, and facial recognition due to their ability to effectively handle spatial relationships and hierarchical patterns within data.

Built With

Setup Instructions

1. Clone the Repository: Copy this repository to your local machine.

2. Download the Dataset: Download the dataset from Kaggle here.

3. Save the Dataset: Save the icml_face_data.csv file into a folder named Data.

4. Load the Dataset: The notebook is configured to load the dataset from the Data folder. Ensure that the file path is correct.

5. Additional Configuration: Additional configuration might be needed based on your environment and setup.

Recommendations

We strongly suggest using Google Colab to execute the notebook and leveraging a TPU environment for running the code, especially if you plan to train the models yourself. Alternatively, you can import the pre-tuned and baseline models from the GitHub repository. The demo function, camera_feed, was originally designed for use with models on a local system. If you want to test the demo, we recommend loading one of the pre-built models into the function and experimenting with it on your own!

Acknowledgments

• Understanding of Convolution Neural Netowrks
• Get Heatmap from CNN (Convolution Neural Network), AKA CAM
• Face Detection in Python Using a Webcam

(back to top)