← Back to VOLUME 13, ISSUE 6, JUNE 2026
This work is licensed under a Creative Commons Attribution 4.0 International License.
TRUSTNET.AI – DETECTION OF FAKE IMAGES AND VIDEOS
Ms. Bhargavi H G, Himani B L, Sandra C, Navya Manosri S, Ananya V
👁 2 views📥 0 downloads
Abstract: The artificial intelligence technology is getting better and better. This has led to the creation of deepfake technologies. These deepfake technologies are causing problems for cybersecurity and digital trust. They are also making it hard to verify identities. There have been a lot of cases of deepfake fraud in North America. In fact, the number of cases has increased by about 1,740%. This is a jump. 40% Of all biometric authentication attacks worldwide are because of deepfakes.
India is also facing a lot of problems with deepfakes. The number of deepfake-related cybercrimes in India has increased by 550% since 2019. This is a concern. It is estimated that the financial losses due to deepfakes will be than ₹70,000 crore in 2025. These losses will be because of identity impersonation, financial fraud and social engineering attacks. We need to find a way to stop these deepfakes. We need a system that can detect deepfakes with accuracy.
There are already some deepfake detection systems. These systems use something called Convolutional Neural Network (CNN)-based architectures. They are trained on datasets that contain about 15,000 images. These systems are good at detecting deepfakes. They can even work in time. However they have some limitations. They rely on single-model architectures. They do not have comparative analysis capabilities. They also do not have user authentication mechanisms.
To solve these problems, we are proposing a system called the Tri-Model Hybrid Deepfake Detection Pipeline (TMHDPP). This system uses deep learning architectures. It combines the features of these architectures to improve deepfake classification performance. The TMHDPP system uses something called InceptionV3 and EfficientNet. It also uses an architecture that combines the convolutional layers of both models. This helps to capture and aggregate -scale spatial features. The system is. Evaluated using a dataset that contains 2,041 training images and 2,041 testing images.
The TMHDPP system is not a detection system. It also has a web-based application. This application has role-based access control. This means that administrators and users have functionalities. Administrators can upload datasets manage users, train models, test models and monitor performance. They can also visualize the results using graphs.
Users can upload images or videos for deepfake analysis. They can get the classification results along with the evaluation metrics. The TMHDPP system is designed to be scalable, accurate and user-centric. It aims to provide a solution to the growing problem of deepfakes.
The TMHDPP system is an improvement over the existing systems. It uses a - model architecture. It has benchmarking capabilities. It also has integrated management features. All these features make the TMHDPP system a powerful tool for detecting deepfakes. We hope that this system will help to reduce the number of deepfake-related cybercrimes. We also hope that it will help to improve cybersecurity and digital trust. The deepfake detection technology is still evolving. We need to keep working on it to make it better. The TMHDPP system is a step in the direction. It has the potential to make a difference in the fight, against deepfakes.
India is also facing a lot of problems with deepfakes. The number of deepfake-related cybercrimes in India has increased by 550% since 2019. This is a concern. It is estimated that the financial losses due to deepfakes will be than ₹70,000 crore in 2025. These losses will be because of identity impersonation, financial fraud and social engineering attacks. We need to find a way to stop these deepfakes. We need a system that can detect deepfakes with accuracy.
There are already some deepfake detection systems. These systems use something called Convolutional Neural Network (CNN)-based architectures. They are trained on datasets that contain about 15,000 images. These systems are good at detecting deepfakes. They can even work in time. However they have some limitations. They rely on single-model architectures. They do not have comparative analysis capabilities. They also do not have user authentication mechanisms.
To solve these problems, we are proposing a system called the Tri-Model Hybrid Deepfake Detection Pipeline (TMHDPP). This system uses deep learning architectures. It combines the features of these architectures to improve deepfake classification performance. The TMHDPP system uses something called InceptionV3 and EfficientNet. It also uses an architecture that combines the convolutional layers of both models. This helps to capture and aggregate -scale spatial features. The system is. Evaluated using a dataset that contains 2,041 training images and 2,041 testing images.
The TMHDPP system is not a detection system. It also has a web-based application. This application has role-based access control. This means that administrators and users have functionalities. Administrators can upload datasets manage users, train models, test models and monitor performance. They can also visualize the results using graphs.
Users can upload images or videos for deepfake analysis. They can get the classification results along with the evaluation metrics. The TMHDPP system is designed to be scalable, accurate and user-centric. It aims to provide a solution to the growing problem of deepfakes.
The TMHDPP system is an improvement over the existing systems. It uses a - model architecture. It has benchmarking capabilities. It also has integrated management features. All these features make the TMHDPP system a powerful tool for detecting deepfakes. We hope that this system will help to reduce the number of deepfake-related cybercrimes. We also hope that it will help to improve cybersecurity and digital trust. The deepfake detection technology is still evolving. We need to keep working on it to make it better. The TMHDPP system is a step in the direction. It has the potential to make a difference in the fight, against deepfakes.
How to Cite:
[1] Ms. Bhargavi H G, Himani B L, Sandra C, Navya Manosri S, Ananya V, “TRUSTNET.AI – DETECTION OF FAKE IMAGES AND VIDEOS,” International Advanced Research Journal in Science, Engineering and Technology (IARJSET), DOI: 10.17148/IARJSET.2026.13631
This work is licensed under a Creative Commons Attribution 4.0 International License.