πŸ“ž +91-7667918914 | βœ‰οΈ ijarcce@gmail.com
International Journal of Advanced Research in Computer and Communication Engineering
International Journal of Advanced Research in Computer and Communication Engineering A monthly Peer-reviewed & Refereed journal
ISSN Online 2278-1021ISSN Print 2319-5940Since 2012
IJARCCE adheres to the suggestive parameters outlined by the University Grants Commission (UGC) for peer-reviewed journals, upholding high standards of research quality, ethical publishing, and academic excellence.
← Back to VOLUME 15, ISSUE 11, NOVEMBER 2026

Training Data Diversity and Distribution Shift in AI-Generated Image Detection: An Empirical Study

E Niranjan, Arushi, Deesha S Chajer, Niveditha S

πŸ‘ 16 viewsπŸ“₯ 2 downloads
Share: 𝕏 f in ✈ βœ‰
Abstract: The proliferation of high-fidelity generative models such as Stable Diffusion, DALL-E, MidJourney, and Gemini has made automated detection of AI-generated images an increasingly important research problem. While many published detectors report high in-domain accuracy, their performance under distribution shift is often overlooked. This paper presents an empirical study of binary real-versus-AI image classifiers, examining how training data composition and architecture choice affect accuracy on images from sources not seen during training. Models trained only on public data (ResNet18 and EfficientNet-B3) reached 53.33% and 61.67% on a 60-image held-out set of images from web-based generators and personal photographs, whereas a ResNet18 trained with an additional 360 self-collected images reached 75.00%, rising to 81.67% after threshold calibration. On unseen Tiny-GenImage samples (n=800) the three models scored between 72.12% and 74.75%, and the data-augmented model scored lower than the public-only ResNet18 on a 2,000- image slice (68.85% vs. 72.60%), indicating a trade-off between the added sources and the original public distribution. Because the two architectures have near-identical parameter counts (11.18M and 10.70M) and the runs differ in training volume and schedule, we draw no conclusion about model capacity. The results suggest that adding even a small amount of data from the target sources can matter more than the choice between these two architectures.

Keywords: AI-generated image detection, deep learning, convolutional neural networks, dataset diversity, transfer learning.

How to Cite:

[1] E Niranjan, Arushi, Deesha S Chajer, Niveditha S, β€œTraining Data Diversity and Distribution Shift in AI-Generated Image Detection: An Empirical Study,” International Journal of Advanced Research in Computer and Communication Engineering (IJARCCE), DOI: 10.17148/IJARCCE.2026.151159

Creative Commons License This work is licensed under a Creative Commons Attribution 4.0 International License.