Comparative Evaluation of Deep Learning Models for Fake Image Detection

TL;DR

This study compares four pretrained CNN models for fake image detection, highlighting their performance, limitations, and the importance of balanced datasets and advanced training techniques.

Contribution

It provides a reproducible baseline comparison of CNN architectures for fake image detection and discusses key challenges like dataset imbalance and model interpretability.

Findings

VGG16 achieved 91% accuracy, outperforming others.

EfficientNetB0 was more sensitive but less reliable on real images.

Limitations include dataset imbalance and overfitting.

Abstract

The growing sophistication of GAN-based image manipulation presents significant challenges for digital forensics. This study compares the performance of four pretrained CNN architectures including VGG16, ResNet50, EfficientNetB0, and XceptionNet for fake image detection using a unified preprocessing and training pipeline. A dataset of real and manipulated images was processed through resizing, normalization, and augmentation to address class imbalance and improve generalization. Models were evaluated using Accuracy, Precision, Recall, F1-score, and ROC-AUC. VGG16 achieved the highest accuracy at 91%, with XceptionNet, ResNet50, and EfficientNetB0 each reaching 90%. EfficientNetB0 showed stronger sensitivity to fake images but reduced reliability on real samples, reflecting imbalance-driven bias. Limitations include dataset imbalance, overfitting, and limited interpretability, which affect cross-domain robustness. The study provides a reproducible baseline and underscores the need for balanced datasets, advanced augmentation, and fairness-aware training to develop reliable fake image detection systems.