Image Splicing Detection using Deep Residual Network

Forgery using images are common nowadays. This may result in misleading the court, changing the mindset of people and defaming an individual. It is the need of the hour to design a tool that can detect forged and authenticated images. Image forgery detection schemes may be active or passive. Tampering detection schemes come into the category of passive or blind image forgery detection schemes. Deep Learning is a technique used to recognize or classify images into multiple class. Images are used as input for the convolutional neural network and processed through various layers for feature extraction and these extracted features are used as a training vector for the classifier model. This paper uses a pre-trained Deep Learning model resnet-50 for feature extraction from CASIA 2.0 dataset and three different classifiers for classification purpose.