F.A.Q. #4
Replies: 9 comments 5 replies
-
|
will there be a need for donated CPU and GPU power during training/back prop or does it need to be trained on a single machine and so a cloud service platform would be used? as unfortunately i don't have any expertise in the fields wanted but would still like to contribute more than just game play. |
Beta Was this translation helpful? Give feedback.
-
|
we haven't quite decided on that yet! If we end up doing a distributed training method then we'll make sure to let everyone know! |
Beta Was this translation helpful? Give feedback.
-
|
For gameplay footage how long should the video be? |
Beta Was this translation helpful? Give feedback.
-
|
for now, PC footage. 5-10s clips are best. |
Beta Was this translation helpful? Give feedback.
-
|
Will it be able to expand to more than just fps games? Also, when will it be available for CS? |
Beta Was this translation helpful? Give feedback.
-
|
Judging by the nature of the tech, I'm not sure what other games besides FPS this would be applicable for, since it primarily reads recoil patterns. What did you have in mind? |
Beta Was this translation helpful? Give feedback.
-
|
Posting this here because it's not specified very clearly on the initial FAQ, but the initial goal is for M&K cheat detection, not controller footage. If you're here to contribute footage from controller gameplay you may have to wait a bit for the Project focus to get to that. |
Beta Was this translation helpful? Give feedback.
-
|
Where should we post clips, and should they still only be Valorant clips. Also are they supposed to look sus or can should there also be random gameplay? |
Beta Was this translation helpful? Give feedback.
-
|
I'm not actually a dev myself (lack the coding experience with visual analysis software) but seems currently they're focused on CSGO as their baseline. Secondly, they do not have any working software as of yet, the team is still putting together resources and people and have begun work, but not actually generated any product yet. I'm not sure where they're are taking clips at this time, or if they are even taking any until they get a baseline for the program to learn from. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for letting me know. |
Beta Was this translation helpful? Give feedback.
-
|
How will you prevent people form tuning their machine learning based cheating algorithms using this project to make them undetectable again? This project was going to happen eventually and if you manage to do it, it will likely achieve some short term success, but philosophically I feel the fight is already lost as the end of this fight you just can't make the difference between the best human performance and an algorithm trying to mimic it. |
Beta Was this translation helpful? Give feedback.
-
|
Why is your WALDO not licensed yet? If licensed then what license is it and why was it not included? |
Beta Was this translation helpful? Give feedback.
-
|
I prefer using insight.gg as gameplay recordings, is this acceptable instead of OBS? And I only play Apex out of all the other games that were listed here |
Beta Was this translation helpful? Give feedback.
-
|
I would love to supply footage when you feel ready for it. Just let us know where you want it uploaded, what type of footage you want, etc, and I'll help. |
Beta Was this translation helpful? Give feedback.
-
Welcome to the Project! Check out the FAQs below before getting started.
Frequently Asked Questions
What is this project's goal?
A deep learning Artificial Intelligence (A.I.) can detect the human behavioral characteristics of a user within a video game. We plan to train an A.I. to understand how humans play video games via a visual machine learning program. Once the program understands how humans play video games based on gameplay footage, we can then feed the it gameplay footage to determine if the player in the footage is receiving assistance from a 3rd party "hack" or "cheat" program.
The first goal to achieve is detection of the most prevalent "closet hack" which is humanized aim-assist. We will build a program that can extrapolate mouse movement data from gameplay footage via 3D imaging and optical flow. Once accurate mouse data can be logged from gameplay footage successfully, we can then begin to train the A.I. what human aim looks like in a video game. This will be done by providing data to the program from gameplay that has been verified as human. With enough data, we will have a trained A.I. program that can tell if aim within any given piece of gameplay footage is human or machine assisted.
Future iterations of the program will include detection of many more forms of closet hacking.
What is the timeline on this project?
This is dependent on how much help we get. Each step of the project provides a new list of problems to tackle. Once we start making progress on the first few steps, we will be able to estimate an accurate timeline for release.
How can I help?
Do you have programming skills? A.I. / Machine learning skills? Research skills? We need your help!
Check out the Projects tab for a list of the features we're currently working on and hop into the welcome discussion thread or the discord channel to start contributing.
If you're post in the welcome discussion includes references to show you're experienced in areas that we need, you will be added to the "devs-only" section of the Discord. We will discuss deeper development topics in that text channel.
Research and creative discussion will be a big part of this project from start to finish. Join the Discord, give yourself the 'Anti-Cheat' role in the #self-rolls channel, and contribute via the "ANTI CHEAT" text channels. https://discord.com/invite/SgpnrUm
If you don't have any special programming skills, stay tuned! We're going to need LOTS of gameplay footage in the near future in order to feed our model training data. Get your OBS hotkeys prepared, gamers.
Beta Was this translation helpful? Give feedback.
All reactions