How Dogs (Both AI generated and real) can help train AI models to spot deepfakes.

Student: Anamitra Abi
Table: 10
Experimentation location: School, Home
Regulated Research (Form 1c): No
Project continuation (Form 7): No

Display board image not available



No additional citations

Additional Project Information

Project website: -- No project website --
Research paper:
Additional Resources: -- No resources provided --

Research Plan:

Questions and Answers

Iniitial project questions

1. What was the major objective of your project and what was your plan to achieve it? 

I wanted to develop an AI model that could successfully recognize the difference between an AI generated image and a real image.

    a. Was that goal the result of any specific situation, experience, or problem you encountered?  

I had read about how deepfakes were a big issue and wanted to make a micro solution that could be utilized to solve a macro issue.

    b. Were you trying to solve a problem, answer a question, or test a hypothesis?

I was trying to answer a question: “Can an AI model recognize the difference between a AI generated image and a reall one?”


2. What were the major tasks you had to perform in order to complete your project?

I had to create data sets of 25 and 25 for the real and AI dogs and 10 dogs (5 and 5) to test. Then I had to develop the model and test it again and again.

    a. For teams, describe what each member worked on.


3. What is new or novel about your project?

    a. Is there some aspect of your project's objective, or how you achieved it that you haven't done before?

I hadn’t seen any projects that tried to solve this issues of deepfakes with a simple solution that could be scaled up in size and intricacy. 

    b. Is your project's objective, or the way you implemented it, different from anything you have seen?

Yes, during my research I saw nothing quite like it.


4. What was the most challenging part of completing your project?

Testing the model was not only tedious but the platform of Teachable Machine posed several problems for me throughout the project.

   a. What problems did you encounter, and how did you overcome them?

At one point, I lost all the data  had and had to recreate the data sets because Teachable Machine had lost all the information on which dog was AI and which dog was real.

   b. What did you learn from overcoming these problems?

I learned ot consolidate my data sets into different folders, which made the next couple of times the problem occurred muche easier to deal with.


5. If you were going to do this project again, are there any things you would you do differently the next time?

I would like to try the same experiment again with a larger sample size, I feel that could provide me with even more conclusive data. 



6. Did working on this project give you any ideas for other projects? 

I would also like to try a similar experiment with human faces in order to understand how AI interprets human faces.


7. How did COVID-19 affect the completion of your project?

My project was based off of my access to a computer so it did not affect me at all.