Transfer Learning for Detecting Fake Images that Resulted from Turkey Earthquake

Background: On February 6, 2023, a devastating 7.8 magnitude earthquake struck southern and central Turkey, as well as northern and western Syria, causing widespread destruction and loss of life. It was reported as the deadliest earthquake in the history of the region, with thousands of casualties and significant damage to physical and digital infrastructures. Objectives: After the devastation caused by the earthquake, some news articles and social media reports resorted to spreading false news by twisting the facts and manipulating images transferred by the media. This fake news may affect the quality of business reports prepared for measuring the earthquake damage. Therefore, this study comes to attract people’s attention to some cosmetic aesthetics that most people may find interesting rather than viewing the real value of the loss. Methods: In this chapter, we resorted to a dataset collected from the Getty Images platform where a large group of real and fake images are available. This dataset supports us in drawing attention to the erasures and injustices that have occurred by doing some plastic facts entitled “fake news.” Thereby, our work opens the door to emphasize how this information is being erased, manipulated, and distorted by the media. We used a convolutional neural network (CNN), a widely applied deep learning technique for determining whether the image is real or fake. Results: After applying the classification task to Turkey earthquake images by using deep learning, we become able to detect fake news with a competitive performance that demonstrates the feasibility of our approach.