nk9487 / ad470-practicum

AD470-PRACTICUM

Real Or AI competition This repo is for the Real Or AI Kaggle competition. The purpose is to create a model that can classify AI & real generated images Link:https://www.kaggle.com/competitions/real-or-ai/overview

Participants in this exciting competition will be challenged to develop a model that can accurately differentiate between images generated by artificial intelligence and those captured in reality. This task is essential in various fields, including computer vision, security, and art, where distinguishing between the two is critical. In the field of computer vision, the ability to distinguish between AI-generated and real images is vital for many applications, such as detecting deepfake videos, identifying fraudulent product images, and verifying the authenticity of medical imaging. Accurately classifying these images can help prevent malicious actors from deceiving individuals or organizations through the use of AI-generated images. In the realm of security, the capability to differentiate between real and AI-generated images is essential for thwarting fraud and preventing identity theft. Criminals can create false IDs or passports using AI-generated images, making it crucial to differentiate between the genuine and fake ones. Developing accurate models to differentiate between the two types of images can play a critical role in ensuring security and protecting individuals' identities. In the art world, distinguishing between real and AI-generated images is necessary to preserve the value and authenticity of artwork. The rise of AI-generated art has made it increasingly challenging to differentiate between works created by humans and those made by machines. Creating models that can accurately classify these images can help ensure that the value and authenticity of artwork are maintained, while also allowing for new forms of artistic expression. In conclusion, the Real or AI competition offers an exciting opportunity to develop models that can accurately differentiate between images generated by AI and those captured in reality. This task is crucial in various fields, including computer vision, security, and art. Participants will have the chance to make a significant impact in these fields by developing accurate models to distinguish between AI-generated and real images, thereby advancing the state of the art in image classification.

ad470-practicum's Issues

Clean Up and Troubleshoot Code to improve accuracy.

Review code and modify to try and identify ways to improve accuracy

Train Petal to the metal competition model

Work on code to train model

Identify additional Kaggle competition projects can expand into

Find a way to measure the accuracy of result based on test data and try to tian more data.

Research TPUs and how to implement

Add training data set to model to increase accuracy