AimAide for CSGO ~~and soon CS2~~

*UPDATE! <br>as long as raw input can not be disabled in CS2 the mouse mover does not work.<br> models seem to work tho

External realtime object detection-based aim aiding powered by <b>YOLOv8</b>, <b>CUDA</b> and <b>TensorRT</b><br> Twitter for further development updates

<img src="/docs/header_cts.jpg"><br> Video Demo here <br>

<h3>Latest changes/additions</h3> <br> <b>13/09/23</b><br> fixed initial TensorRT engine building when starting for the first time from webui launcher<br> changed ultralytics version to required 8.0.58, cause there are some issues with newer versions <br> <br> 25/07/23<br> -added a webui launcher for comfortably changing settings<br> just run launcher.py from your command line <br> <hr> <img src="/docs/launcher.png"> <br> <hr> 16/05/23<br> -new model which is optimized on heads on mirage (high headshot rate, --minconf 0.6 recommended)<br> models/yolov8s_csgo_mirage-320-v62-pal-gen-bg-head.pt<br> -added flickieness argument to control how fast the mouse mover should flick to target <br><br> 09/05/23 - bug in d3d_np grabber fixed (mixed up color channels), code improvements, removed engines from repo (engines will built locally),<br> d3d_gpu is disabled and needs to be rewritten<br> 16/04/23 - engine builder added to circumvent TensorRT incompatibilities <br>(by https://github.com/triple-Mu/YOLOv8-TensorRT)<br> 15/04/23 - introduced 320x320 input models which drastically increase fps with YOLO and TensorRT<br> <h3>Supported Maps</h3> * Mirage <h3>Road Map</h3> Models for CS2 and support for additional maps<br> Human-like aim methods (like windmouse or ai-based) <h3>Features</h3> YOLOv8 Models trained on mirage with various CT and T agents (body and head).<br> Simple smooth linear mouse mover locking onto target closest to crosshair.<br> <h3>Hardware Requirements</h3> To get this to work the detector has to run at 30fps at least.<br> NVIDIA GTX1070 runs at 30fps on a 640x640 model or 60fps on a 320x320 model with TensorRT.<br> NVIDIA RTX4090 should max out at ~120fps on a 640x640 model. (also TensorRT)<br> <h3>Installation</h3> 1) NVIDIA CUDA Toolkit >= 11.7<br> 2) Python 3.10.6 environment<br> 3) Corresponding PyTorch CUDA Package -> https://pytorch.org/get-started/locally/<br> 4) pip install -r requirements.txt<br><br> <b>Optional but recommended:</b><br> 5) NVIDIA TensorRT >= 8.4 -> https://developer.nvidia.com/tensorrt -> https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html<br> <br> Speedup for bigger models with TensorRT is significant.<br> <s>Thats why all models bigger than medium will only be released as TensorRT engines</s>.<br> Models will be released as YOLO weights and locally built as TensorRT engines on first start up.<br> This is due to TensorRT version incompatibilities.<br> <img src="/docs/TensorRT_Speedup.png"> <h3>Usage</h3> I) Disable windows mouse acceleration<br> II) Disable raw input in CSGO<br> III) Cap max_fps in CSGO at your native display refresh rate<br> <br> 1) Run either run_tensorrt.py or run_yolo.py<br> 2) Selective detection can be activated by running with argument <b>-side 'your side'</b> (t, ct or dm for detecting all)<br> If you want to change the detection mode while the script is running, simply write 't', 'ct' or 'dm' into the console and hit enter<br><br> <img src="/docs/side_switch.png"><br> 3) Depending on your hardware choose from 3 different models (nano, small, medium)<br> nano (highest framerate, lowest detection performance),<br> medium (lowest framerate, best decetion performance)<br> 4) Run in benchmark mode first to see what framerate you get (over 60fps increase sensitivity mode)<br> 5) Adjust mouse sensitivity in CS and/or sensitivity mode of AimAide <h3>Benchmark mode</h3> Run run_tensorrt.py or run_yolo.py with argument <b>--benchmark</b> to start in benchmark mode.<br> This is going to run the detector in view-only- and detect-all mode for 300 iterations.<br> Switch to CSGO and run/look around. At the end the average fps of the detector during that time will be displayed. <br><br> <img src="/docs/benchmark_mode1.png"> <h3>Arguments<h3>

| arg | default | Description | | ---- | --- | --- | | _{--input_size} | ₃₂₀ | _{dimension of the input image for the detector} | | _--grabber | _'win32' | _{select screen grabber (win32, d3d_gpu, d3d_np)} | | _--model | _{models/yolov8s_csgo_mirage-320-v41-al-gen-bg}| _{selected engine (TensorRT) or weights (YOLOv8)}|
| _--side | _'dm' | _{which side your are on, 'ct', 't' or 'dm' (deathmatch)} | | _--minconf | _0.75 | _{minimum detection confidence} |
| _{--sensitivity} | ₁ | _{sensitivity mode, increase when having a high framerate or chaotic aim} | | _{--flickieness} |₄ | _{how flicky the mouse mover behaves (4 is slow, 16 is very flicky)} | | _--visualize |_False | _{show live detector output in a new window} | | _{--view_only} |_False | _{run in view only mode (disarmed)} | | _--benchmark | _False | _{launch benchmark mode} | | _{--no_engine_check} | _False | _{skips engine checking and building (run_tensorrt.py only)} |

<h3>FAQ</h3> Q: Why does AimAide seem to be stuck on launch?<br> A: This is a known issue with the YOLO class, run your command line as administrator.<br> <br> Q: Why is the aiming is so chaotic and unnatural?<br> A: Probably due to high detector framerate. Increase the sensitivity mode by running with arg -sensitivity (default is 1)<br> <br> Q: Why is the aiming is so slow and laggy?<br> A: Probably due to low detector framerate. Run benchmark mode and check if you get an average fps of at least 30 while being ingame.<br> <br><br> Feel free to fork this repo and/or use the models for your own projects. :)