The rapid advancement of digital technologies has made image manipulation more accessible and increasingly sophisticated. Among the most common forms of forgery is image splicing, which involves combining parts of two or more images into a single image. This practice is often executed with malicious intent, such as spreading misinformation or altering evidence. Traditional detection methods relying on handcrafted features have proven inadequate in addressing the seamless blending achieved through modern editing tools. This study suggests a hybrid architecture that combines Support Vector Machine (SVM) for classification with the ResNet50 deep learning model for feature extraction in order to overcome these difficulties. Leveraging ResNet50’s capability to capture intricate spatial features and SVM’s effectiveness in classification, the framework demonstrates robust performance in distinguishing authentic images from spliced ones. When evaluated on the CASIA v2.0 dataset, the model achieves 92.05% accuracy even under challenging conditions like compression and resizing. This work underscores the potential of combining deep learning and machine learning techniques to enhance digital forensics and uphold the integrity of visual content.