Deep Fake Detection Challenge

(David Pfahler) #1

https://deepfakedetectionchallenge.ai

I am very much looking forward to this. What do you think?

9 Likes

#2

Sounds very interesting. I think we can leverage some of the looking into the model in Part 2 to see where the mistake occurs and potential ways of fixing it

0 Likes

#3

Seems quite interesting! I hope it is framed to be logic intensive and not compute-intensive which only few can participate vigorously.

0 Likes

(Nathan Hubens) #4

I’m really interested in the challenge and I’ll probably participate :slight_smile:

0 Likes

(Elena) #5

Sounds like an interesting topic/challenge! I’ll keep watching this thread for some updates :slight_smile:

0 Likes

(RobG) #6

Looks like the challenge itself starts in December, launched at NeurIPS.

2 Likes

(RobG) #7

Initial dataset has been released

1 Like

(RobG) #8

Now on Kaggle …

2 Likes

(Dave) #9

Yes, I am interested in what ever is being defined as “deepfake” since its a very much something I’d like to better understand. The kaggle/facebook task is totally worth taking a look IMHO, nice prize money up for grabs

Google also released a deepfake dataset a couple of months ago https://ai.googleblog.com/2019/09/contributing-data-to-deepfake-detection.html

1 Like

(Aman Arora) #10

Yes, this sounds like a great challenge to work on.

Here’s FaceForensics++ and here is another research paper for DeepFake detection.

Wondering if anyone is interested in forming a team to get started on this competition and perhaps try and implement ideas mentioned in the research papers?

1 Like

(Sanyam Bhutani) #11

Given the constraints for the notebooks and the dataset sizes mentioned-it should be a very interesting one! :tea:

0 Likes

#12

This seems like it might be pretty compute-intensive, which is pretty disappointing.

0 Likes

(Sanyam Bhutani) #13

@ilovescience https://www.kaggle.com/c/deepfake-detection-challenge/discussion/121279 :wink:

1 Like

(Nathan Hubens) #14

I’m interested to make a team ! I have already started implementing some papers like this one. This will be a very interesting challenge !

1 Like

(Aman Arora) #15

Ah yes! That’s great! I am pretty sure that I read somewhere that MesoNet and Capsule methods have shown best results in DeepFake detection, so that’s a great start!

0 Likes

#16

Can I join?

0 Likes

(Nathan Hubens) #17

Sure !

1 Like

#18

Not sure if this works but AWS gives thousands in credits for this comp. I emailed them and awaiting their response .

All participants will be able to request a minimum of $1,000 in AWS credits to get started, with additional awards granted in quantities of up to $10,000 as entries demonstrate viability or success in detecting deepfakes. Participants can visit www.aws.amazon.com/aws-ml-research-awards to learn more and request AWS credits.

1 Like

#19

can you let me know what they say please?

1 Like

#20

Can I join too? Sent you a linkedin inv!

1 Like