Various softwares like as Adobe Photoshop, Canva, Kine Master, Quicksort, and others are used to create fake images nowadays. We're concentrating on images that have been forged using image splicing techniques for this project. Using a convolutional neural network and semantic segmentation, image forgeries are discovered and detected. A deep convolutional neural network transfer learning approach was used to teach the algorithm with two classes. The suggested system comprises two phases: detection and localization. The two algorithms used in the proposed system are VGG16 and U-Net.Utilize deep convolutional neural networks to teach the algorithm with two classes. VGG16 is a 16-layer CNN architecture that is used to classify whether image pixels have any forgery or not. UNet and VGG16 are used separately as two different methods to train semantic segmentation to locate forged pixels on images that are classified with colour pixel labels. VGG16 performs well with good results out of the two methods utilised for localization.

ConvolutionalNeuralNetwork(CNN),Pixels , Image Forgery , Image Splicing,SVM