In recent years, diffusion models have become a powerful tool in generative
artificial intelligence, capable of creating highly realistic images, often resem-
bling real-world objects and scenes, making them nearly indistinguishable
from genuine ones. These synthetic images, also known as deepfakes, present
a significant challenge in differentiating between authentic and artificially
generated content. This diploma thesis addresses this issue through participa-
tion in a competition aimed at detecting images generated by state-of-the-art
diffusion models. In this work, we compared various solutions and attempted
to improve existing approaches to detecting these deepfake images. By eval-
uating different methodologies, we explored how effectively real images can
be distinguished from those generated by diffusion models. This work not
only contributes to the ongoing research in this field but also highlights the
challenges and potential strategies for enhancing deepfake detection.
|
