Requirements

• Ensure you have Python version 3.11.6 or higher installed.
• Install the ultralytics package:

  pip install ultralytics -U

Repository Unpacking

• Download the repository (click the green "Code" button and then "Download ZIP"), then extract the archive to a convenient location so that everything is in one folder.

Annotating Images / Creating a Dataset / Tips

• There are many programs and services for annotating images. If you're doing this for the first time, it's recommended to use LabelImg.

Installing and Setting up LabelImg

1. Download LabelImg or LabelImg_Next.
2. Extract the program and move the predefined_classes.txt file (located in the repository) to the labelimg/data folder.

Repository Structure with Annotated Image Examples

The repository already contains examples of annotated images in the ai_aimbot_train/datasets/game folder:

• images – contains images for model training.
• labels – contains annotations for the corresponding images. These are text files with the coordinates of bounding boxes and object class IDs.
• val – used to store images and annotations for the validation dataset. The validation set is needed to check the model's performance during training and prevent overfitting.

Annotating Your Own Data

1. Open LabelImg and click Open Dir, selecting the path to the images for model training (ai_aimbot_train/datasets/game/images).
2. Click Change Save Dir to choose where the annotations will be saved (ai_aimbot_train/datasets/game/labels).
3. (Optional) Familiarize yourself with the LabelImg usage guide.
4. (Optional) Explore tips for achieving the best training results.

Recommendations for Creating a Quality Dataset

• Annotate between 500 to 2500 images.
• Include empty images that lack players, weapons, fire, and other objects. For example, add lots of images with trees, chairs, grass, human-like objects, and empty game locations.
• The more complex a game looks for AI (e.g., CS2 is more formalized than Battlefield 2042), the more data you'll need for model training (at least 5000-10000 images).
• Image resolution can vary from 100x100 to 4K.
• Ready-made datasets can be found here.
• Don't forget to add images and annotated files to the val folder. If you have 1000 training images, add about 10 validation images.

Setting up the game.yaml File

• The file specifies the following parameters:

  path: game  # Your dataset name
  train: images  # Folder with training images
  val: val  # Folder with validation images
  test:  # Folder with test images (can be empty)

• To find Ultralytics settings file, enter the command:

  yolo settings

Model Training

• (Optional) Detailed information about training can be found here.

• After annotating the dataset, navigate to the ai_aimbot_train folder using the command:

  cd ai_aimbot_train

• Choose a pre-trained model. The options are:

  • yolo11n or yolo12n – the fastest and least resource-intensive.
  • yolo11s or yolo12s – fast, slightly smarter but more resource-demanding.
  • yolo11m or yolo12m – optimized for real-time, requires a powerful GPU (e.g., RTX 2060 or better).
  • yolo11l or yolo12l and yolo11x – most intelligent and resource-intensive, not suitable for most tasks.

• For example, choose yolo11n:

  model = yolo11n.pt

• Select the image size for the model. The lower the resolution, the faster the training and the fewer objects the model can detect:

  • Possible options: 160, 320, 480, 640. Choose 320.

• Determine the number of training epochs. A larger dataset requires more epochs, but don't overdo it, as too many epochs can lead to overfitting. Assume we set 40 epochs.

• (Optional) All training parameters are described in detail here.

• Start training with the command:

  yolo detect train data=game.yaml model=yolo11n.pt epochs=40 batch=-1

• After successful training, navigate to the model weights folder:

  ai_aimbot_train/runs/detect/train/weights
  

  You'll see two files:

  • best.pt – the file with the best model weights.
  • last.pt – checkpoint of the last epoch. If training was interrupted, you can resume from this file:

    yolo train resume model=ai_aimbot_train/runs/detect/train/weights/last.pt

Model Testing

• Create a test.py file with the following content:

    import cv2
  
    from ultralytics import YOLO
    
    # Load the YOLO model
    model = YOLO("best.pt")
    
    # Open the video file
    video_path = "path/to/your/video/file.mp4"
    cap = cv2.VideoCapture(video_path)
    
    # Loop through the video frames
    while cap.isOpened():
    	# Read a frame from the video
    	success, frame = cap.read()
    
    	if success:
    		# Run YOLO inference on the frame
    		results = model(frame)
    
    		# Visualize the results on the frame
    		annotated_frame = results[0].plot()
    
    		# Display the annotated frame
    		cv2.imshow("YOLO Inference", annotated_frame)
    
    		# Break the loop if 'q' is pressed
    		if cv2.waitKey(1) & 0xFF == ord("q"):
    			break
    	else:
    		# Break the loop if the end of the video is reached
    		break
    
    # Release the video capture object and close the display window
    cap.release()
    cv2.destroyAllWindows()

• Run:

  python test.py

Fine-Tuning the Sunxds Model to Avoid False Detections or Improve Detection of Other Classes

1. Selecting a Pre-trained Sunxds Model

   • Choose a model that is most similar to your task in terms of version and image size (640, 480, 320 or 160).

2. Preparing the Dataset

   • Prepare the dataset with new annotations as specified in this repository. Ensure that the new annotations follow the YOLO format.
   • If you want to eliminate false detections, add images with false detections to the dataset and do not annotate anything on them.
   • If, for some reason, there are poor or no detections of players in a particular game, add annotated images with these players to the dataset.
   • Don't be lazy; add more images to the dataset, and try to review the entire dataset again for errors after annotating it.

3. Running the Fine-Tuning

   • Execute the following command:

     yolo detect train data=game.yaml model=sunxds_0.7.5.pt epochs=40

Export model to run model in sunone_aimbot_cpp:

• Export model to onnx format with dynamic shape:

   yolo export model=best.pt format=onnx dynamic=true simplify=true

• Select .onnx model in AI tab.