/ai-or-human

Python AI image generated detector built using Streamlit and using Hugging Face Model API

Primary LanguagePythonMIT LicenseMIT

AI or Human

Python AI image generated detector built using Streamlit and using Hugging Face Model API

Our Team

Our Process

Application Methodology

Process

Final Report

Prepared for

Cevi Herdian. B.Sc., M.Sc.

Prepared by

Introduction

Andrew, Feri, Eric, and Janice are 6th-semester students majoring in Computer Science at Universitas Bunda Mulia. We created a project titled "AI or Human" as part of our Python Web Framework Programming course final exam. The goal of this project is to detect and determine whether an image is AI generated image or a human made image. We use inference API from Hugging Face for our model and Streamlit as a part of our tech stack because Streamlit was enough to finish this project. We hope that this project could make an impact and benefit the readers.

Executive Summary

The "AI or Human" project aims to detect and determine whether an image is an AI-generated image or a human-generated image.

This project will be implemented in two phases. The first phase is to design the interface for the web application. The second phase focuses on building coding applications using APIs from Hugging Face for the model and Streamlit.

The success that is expected and what is measured from this project is how many input photos are successfully detected according to the classification of the model used.

The project is expected to take three months to complete, with a total budget of $100.

Project Details

DATE TASK
March 2 Project Proposal
March 16 Mock up Design
March 17 Project Proposal Presentation
April 24 Development
May 1 Testing
May 3 Deployement
May 5 Project Closing

Start-Up Cost

The table below presents the expenses incurred during the application development.

DESCRIPTION AMOUNT PERCENTAGE OF TOTAL
Meeting expenses $10 15%
References $22 33%
App deployment $20 30%
Other expenses $15 22%
Total Start-Up Cost $67 100%

Conclusion

AI or human estimates the percentage of AI (Artificial Intelligence) contributions in an image, the higher the value of AI percentage indicates the image is more likely to be generated by AI, and vice versa. Thereby, this app is built to differentiate an image made by AI or humans concerning an AI-generated image being contested in a human image-related art competition.