Deepfake Videos and Images Detector using DL

[anushka29github]

Deep Learning Simplified Repository (Proposing new issue)

:red_circle: Project Title : Deepfake Detection : :red_circle: Aim : To develop and compare multiple machine learning models to detect deepfake videos/images.: :red_circle: **Dataset ** : https://www.kaggle.com/datasets/manjilkarki/deepfake-and-real-images : :red_circle: Approach :

1. CNN
2. CapsNets

---

📍 Follow the Guidelines to Contribute in the Project :

• You need to create a separate folder named as the Project Title.
• Inside that folder, there will be four main components.
  • Images - To store the required images.
  • Dataset - To store the dataset or, information/source about the dataset.
  • Model - To store the machine learning model you've created using the dataset.
  • requirements.txt - This file will contain the required packages/libraries to run the project in other machines.
• Inside the Model folder, the README.md file must be filled up properly, with proper visualizations and conclusions.

---

:red_circle::yellow_circle: Points to Note :

• The issues will be assigned on a first come first serve basis, 1 Issue == 1 PR.
• "Issue Title" and "PR Title should be the same. Include issue number along with it.
• Follow Contributing Guidelines & Code of Conduct before start Contributing.

---

:white_check_mark: To be Mentioned while taking the issue :

• Full name : Anushka Ray
• GitHub Profile Link : https://github.com/anushka29github
• Email ID : [email protected]
• Participant ID (if applicable):
• Approach for this Project : CNN
• What is your participant role? (Mention the Open Source program) GSSoC'24 Contributor

---

Happy Contributing 🚀

All the best. Enjoy your open source journey ahead. 😎

[github-actions[bot]]

Thank you for creating this issue! We'll look into it as soon as possible. Your contributions are highly appreciated! 😊

[abhisheks008]

Hi @anushka29github need to implement 3 models at least for any project.

[zul132]

My Approach:

1. Convolutional Neural Networks (CNNs) can be used to detect whether an image is real or a deepfake by analyzing subtle pixel-level inconsistencies that are often imperceptible to the human eye. In deepfake detection, a CNN is trained on large datasets containing both real and manipulated (deepfake) images. Through convolutional layers, the model learns to identify unique artifacts, such as irregular patterns in textures, lighting, facial edges, or blurring introduced during image synthesis.

2. Capsule Networks (CapsNets) can be effective in detecting deepfake images by capturing spatial relationships and part-to-whole hierarchies more accurately than traditional CNNs. Deepfake algorithms often struggle to maintain consistent patterns across facial regions, such as eyes, mouth, or lighting variations. CapsNets, with their ability to analyze the orientation, position, and pose of features, can identify these subtle inconsistencies. Moreover, the dynamic routing mechanism in CapsNets helps focus on relevant features even if parts of the image are altered or distorted. This makes them well-suited for distinguishing real images from manipulated ones.

3. The Xception model is well-suited for deepfake image detection due to its ability to capture fine-grained artifacts using depthwise separable convolutions. Pre-trained on large datasets, Xception can be easily fine-tuned for detecting deepfake images by adding custom layers. It excels at identifying subtle inconsistencies in facial regions, such as unnatural textures or misaligned features, which are common in manipulated images.

Dataset: https://www.kaggle.com/datasets/manjilkarki/deepfake-and-real-images (its the same dataset that was specified by the issue author)

@abhisheks008 Can you assign me this issue under GSSoC-Ext'2024?

[abhisheks008]

Assigned @zul132

[zul132]

@abhisheks008 Hey I accidently created a PR from my project's branch to the main of my forked repo. I have closed that PR now and created a new PR to your repository's main branch. I don't think this will cause any issues now but I wanted to inform you just in case.

[github-actions[bot]]

Hello @zul132! Your issue #853 has been closed. Thank you for your contribution!