With the increasing performance accuracy of computer software systems, the realistic appearance of computer-generated images (CGI) and deepfakes often leads to assuming them as authentic images.
Researchers at the Changsha University of Science and Technology and Hunan University in Hunan, China, have recently developed an image source pipeline forensics method based on convolutional neural networks (CNN) to automate the distinction between natural images and CGI. The work announced in the International Journal of Autonomous and Adaptive Communications Systems describes that the CNN-based model is fine-tuned using a database of 10000 images.
The team used Inception-v3 as the primary network and adopted pre-trained model parameters in ImageNet. Two fully connected Softmax classifiers replace Inception-v3’s top-level classification layer. The new network model was then constructed using transfer learning.
The evaluation results show that the new model can effectively distinguish between natural and artificial images. This model can also be used with JPEG images, which suffer from compression and scaling artifacts, and effects of post-processing operations that lower their quality and blur the lines between CGI and a digital image.
Identification of CGI and natural images is vital in forensics, politics, and news reports, which have profound implications due to fake, falsified, and fraudulent photos. The model utilizes 2048 dimensions of features in the images extracted by the network for classification to allow the computer to decide the integrity of an image. Currently, the model’s accuracy is as high as 98 percent for certain image types. The team aims to improve the model performance to perform large-scale experimental tests on its accuracy.
Tanushree Shenwai is a consulting intern at MarktechPost. She is currently pursuing her B.Tech from the Indian Institute of Technology(IIT), Bhubaneswar. She is a Data Science enthusiast and has a keen interest in the scope of application of artificial intelligence in various fields. She is passionate about exploring the new advancements in technologies and their real-life application.