Scribd
Upload
24 views2 pages

Fake Image Identification CNN Report

This project develops a Convolutional Neural Network (CNN) model to classify images as real or fake, addressing the challenge posed by digitally manipulated images. The model, trained on a dataset of 'real' and 'fake' images, achieved over 90% accuracy in validation tests. Future work may expand the approach to video deepfake detection and utilize more robust datasets.

Uploaded by

alakuntla937
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
24 views2 pages

Fake Image Identification CNN Report

This project develops a Convolutional Neural Network (CNN) model to classify images as real or fake, addressing the challenge posed by digitally manipulated images. The model, trained on a dataset of 'real' and 'fake' images, achieved over 90% accuracy in validation tests. Future work may expand the approach to video deepfake detection and utilize more robust datasets.

Uploaded by

alakuntla937
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Fake Image Identification Using CNN

Abstract:

This project addresses the increasing threat of digitally manipulated (fake) images using

Convolutional Neural Networks (CNNs). The goal is to develop a deep learning model capable of

classifying images as real or fake with high accuracy.

Introduction:

With the rise of deepfakes and image manipulation tools, identifying authentic images is becoming

increasingly important. This project explores CNN-based binary classification to distinguish real and

fake images.

Dataset Description:

The dataset used consists of two classes: 'real' and 'fake' images. Images were resized to 128x128

and split into training and validation sets. Deepfake datasets such as FaceForensics++ and the

Kaggle DeepFake Detection Challenge were considered.

Model Architecture:

The CNN consists of three convolutional layers with ReLU activation and max pooling, followed by a

fully connected dense layer and a sigmoid output layer for binary classification.

Training:

The model was trained using the Adam optimizer and binary crossentropy loss for 10 epochs,

achieving an accuracy of over 90% on the validation set.

Prediction:
A separate script allows prediction on new images, outputting whether an image is real or fake.

Results:

The model performs well on unseen images and is effective in detecting manipulated content.

Conclusion:

CNNs are a powerful tool for image classification. This project demonstrates their potential in

identifying fake images, and future work can involve extending to video deepfake detection and

more robust datasets.

You might also like