Making a faster AimBot with YOLO. (feat. How to build OpenCV CUDA libraries)

[Music] do here is the question what is the fastest model for objectification yolo man no one can deny that yolo is one of the fastest models for objects detection last time i made a simple aimbot with a tensorflow hub but the problem was it was too slow so this time i'm going to introduce how to use yellow with opencv and make aimbot faster the required python libraries are as follows but if you want to use gpu for opencv you have to build your own library in the last part of this video i will show you how to build opencv for gpu okay let's get started first we need to download the configuration file and widths from the yellow website i'm going to compare tiny yellow and yellow 320. the tiny is super fast but the accuracy is super low and 320 just has average accuracy and speed first load the yellow network model from computer file and wait file if you have built opencv gpu you can enable cuda with this code but if you're not you can just set this code and determine layers here and get a rectangle from a window of counter strike so while getting a frame up counter strike construct a blob from the frame and then perform a forward pass of the yellow object detector here we can change blob size it is usually 416 but we can try many different sizes according to models output has information about bounding boxes class and confidence i set 0.7 as a threshold and class id 0 is for the human scale the bounding box back to the size of image yellow returns the center of detected so i make it as a box here nms boxes does non-maximum suppression to converge duplicated boxes now calculate distance between boxes and crosshair and move the mouse and shoot when we compare it with the tensorflow hub you can see yellow is much faster tiny version shows almost like a rear time performance but as you can see it does not detect well now how to build an opencv gpu library you can just follow this guideline i commented so what we need for this first feature studio and cmake and opencv sources and opencv extra modules and python3 first you need to download opencv sources and extra modules since the extra module is for version 4 4.4.0 so please download opencv version 440 after downloading it extract and make a build folder now you need to generate code with cmake first set the source code as opencv folder and set where to build to your build folder just you mate and hit the configure button and set your platform after configuring is done you need to check these three values and set an extra module pass hit the configure button again after configuration is done now check cuda fast mask and set your cuda architecture you can find your gpu architecture from wikipedia then hit the configure button again after configuration is done hit generate it will take some time and see make this done go to your build folder and open opencv project go to cmake target and build the all build will take around 30 minutes after finish building build install then everything is done but all your building process must be successful so if you successfully build your opencv you can check in python again you can just follow the guideline i commented thanks for watching i hope you enjoyed this video you

Channel: Cheesy AI

Views: 65,090

Keywords: machine learning, deep learning, artificial intelligence, counter strike, opencv, yolo

Id: vQlb0tK1DH0

Length: 5min 33sec (333 seconds)

Published: Sun Mar 07 2021