BUILDING THERMAL HEATING AND COOLING LOAD ESTIMATION USING ARCHITECTURAL DESIGN PARAMETERS

Abstract

With the rapid advancement of artificial intelligence and deep learning technologies, the creation
of highly realistic manipulated media, commonly known as deepfakes, has become increasingly
prevalent. Deepfake images are generated using advanced machine learning techniques that can
alter or replace facial features in images, making it difficult for humans to distinguish between
real and manipulated content.
These technologies can be misused for spreading misinformation, identity theft, cybercrime, and
other malicious activities. Therefore, detecting deepfake content has become an important
research problem in the field of computer vision and digital forensics. This project focuses on
developing an efficient deep learning-based system for detecting deepfake face images using
transfer learning techniques. The proposed system utilizes a pre trained convolutional neural
network model, MobileNetV2, to extract meaningful features from facial images and classify
them as real or fake. Transfer learning is used to leverage the knowledge learned from large
datasets, which helps in improving the model performance and reducing the training time.
The dataset used in this project consists of real and manipulated face images collected from
publicly available sources. The images are first preprocessed by resizing them to a fixed
resolution and normalizing pixel values to make them suitable for model training. The dataset is
then divided into training, validation, and testing sets to evaluate the performance of the model
effectively. During the training process, the MobileNetV2 model acts as a feature extractor, and
additional fully connected layers are added to perform binary classification. The model learns to
identify subtle differences between real and fake images by analyzing patterns, textures, and
facial inconsistencies that are commonly present in deepfake images. Various evaluation metrics
such as accuracy and loss are used to analyze the performance of the model. The experimental
results demonstrate that transfer learning significantly improves the efficiency and accuracy of
the deepfake detection system. The model is capable of identifying manipulated images with
good accuracy and can serve as a useful tool for detecting deepfake content.