YOLO Model Training for Rollercoin Object Detection

Introduction

Today, we’ll train a custom YOLO model to identify objects in Rollercoin game screenshots, focusing on non-human entities like coins and bombs in the “Botcoinclick” mini-game. This custom YOLO tutorial uses YOLOv11 for Rollercoin object detection, starting with downloading a pre-trained model and building from there. Let’s dive into this step-by-step guide YOLO Model Training!

Step 1: Setup and Data Prep

Download YOLO Model

Start by downloading the latest YOLOv11 nano model:

https://github.com/ultralytics/assets/releases/download/v8.3.0/yolo11n.pt

Record and Extract Frames

Record a high-quality video of the Rollercoin game “Botcoinclick.” Extract frames using this script. Save everything in one folder, e.g., RollerCoin.v11i.yolov11, with subfolders:

• yolo11n.pt (downloaded model)
• images (screenshots from video)

Annotate Objects

Use MakeSense.ai to manually label objects (e.g., “coin” and “bomb”). Upload screenshots, draw bounding boxes, and export annotations in YOLO format. Extract the archive into a labels subfolder within your main folder.

Organize Folders

Run this script to split data into train and val subfolders under images and labels.

Step 2: Training the Model

Create Virtual Environment

Set up a new Python environment to avoid conflicts:

python -m venv .venv
.venv\Scripts\activate
pip install ultralytics

Configure data.yaml

Edit data.yaml in your main folder to define paths and classes:

train: F:/PYTHON/BOTS/RollercoinBot-master/RollerCoin.v11i.yolov11/images/train
val: F:/PYTHON/BOTS/RollercoinBot-master/RollerCoin.v11i.yolov11/images/val
nc: 2
names: ['coin', 'bomb']

Set Dataset Path

Update Ultralytics settings at C:/Users/YOURACCOUNT/AppData/Roaming/Ultralytics/settings.json:

"datasets_dir": "F:\\PYTHON\\BOTS\\RollercoinBot-master\\RollerCoin.v11i.yolov11\\"

Train the Model

In your activated environment, run:

yolo train model=F:\PYTHON\BOTS\RollercoinBot-master\RollerCoin.v11i.yolov11\yolo11n.pt data=F:\PYTHON\BOTS\RollercoinBot-master\RollerCoin.v11i.yolov11\data.yaml epochs=50 imgsz=640

After training, the model file (e.g., best.pt) will be in the runs/train folder. Move it to your main folder.

Step 3: Testing the Model

Test your trained model on images or video using these scripts:

• Image test script
• Video test script

Update the script with the path to best.pt.

Conclusion

This YOLO model training guide for Rollercoin object detection equips you to identify coins and bombs in “Botcoinclick” screenshots. Using YOLOv11, free tools like MakeSense.ai, and Python, you’ve built a custom model from scratch. Experiment with more epochs or classes to refine accuracy for your 2025 gaming projects!