BeFR - Deepfake Detection Project

Group Meetings & Activities

9/9/25: Designated Mia as the group leader and discussed interview questions.
9/11/25: Finalized interview questions, plan to have interviews done by next meeting.
9/17/25: Split work for the SRS.
9/18/25: Discussed what software and models to use.
9/21/25: Analyzed project requirements.
9/23/25: Progress report and addressed any concerns.
9/25/25: Finalized SRS.
9/28/25: Collaborated on PowerPoint Presentation.
9/30/25: Fine-tuned report and practiced presentation.
10/17/25: Began working on class, use case, and sequence diagrams.
10/21/25: Continued progress on diagrams and created a low-fidelity prototype.
10/28/25: Finalized stage 2 diagrams.
11/2/25: Went over powerpoint slide and practiced
11/8/25: Presented stage 2
11/11/25: Discussed our plan to implement for stage 3
11/13/25: Split up front end and back end developmet work as well as model training.
11/18/25: Progress report and addresed any concerns
11/19/25: Finalized Deep Fake application and recorded our demo

Validation Results

What is your overall impression of the system? – Our users responded positively to the system and described it as clean, simple, and easy to navigate. Most participants felt the interface was easy to understand and appreciated the extra information included on how to spot deepfakes. Tayla Reitman (27, Technology Consultant) stated “Everything worked the way I expected; upload, click detect, get a result. Very straightforward. I think an explanation on why I got the result I did would also be helpful”. Overall, the users described our system as intuitive and user-friendly. They appreciated that the system avoided overwhelming technical detail.
Do you feel that our system offers an acceptable level of transparency? – In terms of transparency, our users felt that our system provides a reasonable level of clarity by showing the detection result and confidence score. One of our users, Michael Heins (22, Portfolio Analyst), says “The confidence score really gave me a good picture of how sure the system was. It made the result feel a lot more trustworthy”. All of our participants shared similar feedback, that the system is honest and doesn’t leave them feeling confused or unsure. Overall, users agreed that our system communicates effectively and offers the kind of openness they expect from a tool like this.
Do you trust your prediction? Why or why not? – When we asked users if they trusted the systems predictions, most users said they felt confident, especially when the confidence score was high. They appreciated the numerical indicator of certainty. One user, Pamela Wheeless (77, retired businesswoman) told us “Originally, I was hesitant on relying solely on the detector’s prediction, but after using it I changed my mind. The website was easy to use and was very clear with its prediction”. A few people did point out that deepfakes can be super convincing, so they were still a little cautious. Overall, the group felt the system gave believable and helpful results.

BeFR Group Project

Group Members & Roles

Group Meetings & Activities

Upcoming Schedule

Validation Results