Detectd

Detectd is a deepfake detection platform that uses the same technology to detect deepfakes that is used to create them- Artificial Intelligence. Deepfakes are hyper-realistic AI generated/forged media. They are a rising threat to democracies and society in the 21st century. We have developed a sophisticated deep neural network that can analyse videos at a level unmatchable by any other solution, for example - the subject’s consistency in the blinking of an eye, in the video. We study the video as a whole- both on the spatial and temporal dimensions before concluding whether it’s a deepfake or not. Our model is trained on the ImageNet dataset consisting of 14 million images in addition to the sophisticated datasets-Face Forensics++ and Deepfake Detection Challenge Dataset with about 7500 videos- containing both real and real-world deepfake videos. We are in the fight against mis-information warfare by providing Detectd to the institutions and individuals who require it; to protect the idea of democracy, help maintain peaceful societies, safeguard people’s reputation and counter the bad actors from deceiving, manipulating and influencing the public.

Our team consists of 4 members: Atharva Peshkar., Rishita Mishra, Atharva Khedkar and Yash Moharir. We all attend the same college and have known each other for the past 2.5 years. Apart from this project. we've worked together on several other projects. Atharva P. is an IT major undergraduate, Rishita is ETC major undergrad, Atharva K. is an AI major undergrad and Yash is a CSE major undergrad. Atharva P. and Atharva K. specialize in AI and Yash in Web-development. Rishita has an interest in Microsoft Azure and other cloud services.