5 ms·

For shooters, bots or assists that only use vision and normal input (special mouse/keyboard) aren't hard to imagine and I'm positive already exist. I know speci

For shooters, bots or assists that only use vision and normal input (special mouse/keyboard) aren't hard to imagine and I'm positive already exist. I know specifically I read a while back about mice that do recoil correction to improve aim, and things like "click when something that looks like a head is in the crosshairs" would be a weekend project at most for a bored college student. Strategy games may be harder, but still not going to take too long before those exist.

Semaphor 4y ago
Those have been common for decades and are pretty much the lowest bar for detection.
- socialismisok 4y ago
  Using computer vision on the video stream (eg via a capture card) and then sending valid mouse and keyboard inputs most definitely has not been around for decades. It's a mechanism to cheat that has zero binaries running on the host computer.
  - eru 4y ago
    You could do statistical analysis on the mouse and keyboard inputs to try and detect this. (Very similar to how you might detect chess cheaters, actually.)
    - erwincoumans 4y ago
      You can use (the data of) statistical analysis to train a deep neural network to avoid detection. That may also help avoiding chess detection cheaters, so the moves are not perfect anymore.
      - eru 4y ago
        Yes. You can't make cheating impossible, but you can make it harder to pull off.
patmcc 4y ago
It's not hard to make cheating tools, it's hard to run them on the same hardware as the game and not get caught. Chess cheating tools don't need to be on the same hardware, so they're not detectable in the same ways.
- krageon 4y ago
  It is not hard to run cheats for basically any game that uses video on another machine. Just a little more expensive (i.e. it is harder for someone in high school, but not anyone over 20).
  - Orothrim 4y ago
    Due to the need for real-time input that's just not true.
    - bombcar 4y ago
      There are/have been elaborate cheating setups for online shooters that involved a second computer using the VGA or HDMI video stream. It’s doable to get around kernel detection tools.
      - mminer237 4y ago
        Splitting the video stream to a second computer to run real-time video analysis and then send back commands through an emulated mouse may be doable, but it's certainly harder for anyone to do than copying chess moves on your phone.
        bombcar 4y ago
        Certainly! But it does show the amount of effort people will go to cheat - though lots of the "online shooter" cheating is people selling cheating tools/kits to idiots who want to win (basically turning games into pay to win). The number of people who can build that kind of system from scratch is small. But anyone can cheat at chess online.
rowanG077 4y ago
I doubt that since a weekend project. You need to analyze a video stream of hundreds fps in real time with very low latency requirments. You'd probably need an FPGA to do that. Custom training an AI to only trigger on heads.
- rippercushions 4y ago
  No, it's trivial these days, and this very task is one of the introductory tutorials to Google's Coral AI Tensorflow boards: https://coral.ai/models/object-detection/ https://coral.ai/models/object-detection/ Which you can buy as a USB stick for $59: https://coral.ai/products/accelerator/ https://coral.ai/products/accelerator/ Sure, it's a little more complex if you want to train on specific heads as opposed to a generic model, but not hugely so. The biggest challenge would likely be downsampling the video stream into something the model can process quickly enough.
  - rowanG077 4y ago
    That's a generic AI. I would be very surprised you could get it to work accurately for games and hyper specific the heads of (possibly non-human)models. The box you see in the marketing material is completely useless for this use case. Besides how do you get the data to your standard second PC, scale the input the USB thing expects. Send it over USB and back, and then you have to send it to a custom mouse that receives the click command over bluetooth? I don't think 7ms just for the inference step is going to cut it. A weekend project were you have to design an ultra low-latency software + hardware stack, a custom mouse and train a custom AI. I would be incredible if anyone manages this in a weekend.
    - barrkel 4y ago
      7ms isn't going to be the bar. Human reaction times are in the 150+ms range, you have 20x your budget. I agree it's still tough though, most I/O paths on generic hardware will have buffers at every step that you'll have to fight to eliminate.
      - 4y ago
perth 4y ago
Funny thing, there are already YOLOv3 derived aimbots, and a lot of anti cheat now scans pci-e cards to try to detect them. They intercept video input and act as mouse and keyboard devices, acting almost completely independent of the host computer.
- kadoban 4y ago
  Yeah that's not far at all from just running as a separate machine, either grabbing the hdmi output or going full webcam pointed at the screen.

Semaphor 4y ago

Those have been common for decades and are pretty much the lowest bar for detection.

socialismisok 4y ago
Using computer vision on the video stream (eg via a capture card) and then sending valid mouse and keyboard inputs most definitely has not been around for decades. It's a mechanism to cheat that has zero binaries running on the host computer.
- eru 4y ago
  You could do statistical analysis on the mouse and keyboard inputs to try and detect this. (Very similar to how you might detect chess cheaters, actually.)
  - erwincoumans 4y ago
    You can use (the data of) statistical analysis to train a deep neural network to avoid detection. That may also help avoiding chess detection cheaters, so the moves are not perfect anymore.
    - eru 4y ago
      Yes. You can't make cheating impossible, but you can make it harder to pull off.

patmcc 4y ago

It's not hard to make cheating tools, it's hard to run them on the same hardware as the game and not get caught. Chess cheating tools don't need to be on the same hardware, so they're not detectable in the same ways.

krageon 4y ago
It is not hard to run cheats for basically any game that uses video on another machine. Just a little more expensive (i.e. it is harder for someone in high school, but not anyone over 20).
- Orothrim 4y ago
  Due to the need for real-time input that's just not true.
  - bombcar 4y ago
    There are/have been elaborate cheating setups for online shooters that involved a second computer using the VGA or HDMI video stream. It’s doable to get around kernel detection tools.
    - mminer237 4y ago
      Splitting the video stream to a second computer to run real-time video analysis and then send back commands through an emulated mouse may be doable, but it's certainly harder for anyone to do than copying chess moves on your phone.
      - bombcar 4y ago
        Certainly! But it does show the amount of effort people will go to cheat - though lots of the "online shooter" cheating is people selling cheating tools/kits to idiots who want to win (basically turning games into pay to win). The number of people who can build that kind of system from scratch is small. But anyone can cheat at chess online.

rowanG077 4y ago

I doubt that since a weekend project. You need to analyze a video stream of hundreds fps in real time with very low latency requirments. You'd probably need an FPGA to do that. Custom training an AI to only trigger on heads.

rippercushions 4y ago
No, it's trivial these days, and this very task is one of the introductory tutorials to Google's Coral AI Tensorflow boards: https://coral.ai/models/object-detection/ https://coral.ai/models/object-detection/ Which you can buy as a USB stick for $59: https://coral.ai/products/accelerator/ https://coral.ai/products/accelerator/ Sure, it's a little more complex if you want to train on specific heads as opposed to a generic model, but not hugely so. The biggest challenge would likely be downsampling the video stream into something the model can process quickly enough.
- rowanG077 4y ago
  That's a generic AI. I would be very surprised you could get it to work accurately for games and hyper specific the heads of (possibly non-human)models. The box you see in the marketing material is completely useless for this use case. Besides how do you get the data to your standard second PC, scale the input the USB thing expects. Send it over USB and back, and then you have to send it to a custom mouse that receives the click command over bluetooth? I don't think 7ms just for the inference step is going to cut it. A weekend project were you have to design an ultra low-latency software + hardware stack, a custom mouse and train a custom AI. I would be incredible if anyone manages this in a weekend.
  - barrkel 4y ago
    7ms isn't going to be the bar. Human reaction times are in the 150+ms range, you have 20x your budget. I agree it's still tough though, most I/O paths on generic hardware will have buffers at every step that you'll have to fight to eliminate.
    - 4y ago

perth 4y ago

Funny thing, there are already YOLOv3 derived aimbots, and a lot of anti cheat now scans pci-e cards to try to detect them. They intercept video input and act as mouse and keyboard devices, acting almost completely independent of the host computer.

kadoban 4y ago
Yeah that's not far at all from just running as a separate machine, either grabbing the hdmi output or going full webcam pointed at the screen.