YOLO Custom Model for Object Detection
This repository contains a specially trained YOLO model designed for the epic moment UI detection task in the mobile legend (MLBB) video game. The model was trained using a public dataset without including personal information. The pre-trained model is YOLOv8m using a 6 Core 12 thread CPU device and NVIDIA RTX 3090 24GB GDDR6X.
Model Details
- Architecture: YOLO
- Framework: PyTorch
- Performance: mAP@50, precision, recall, etc.
- Model File:
best.pt
Getting Started
Installation
Ensure you have the following dependencies installed:
pip install ultralytics
Usage
To use the model for inference, follow these steps:
from ultralytics import YOLO
# Load the model
model = YOLO("best.pt")
# Perform inference on an image
results = model("path/to/your/image.jpg")
# Display or save results
results.show() # Show predictions
results.save("output/") # Save predictions to output folder
Input Format
- Image: Any standard image format (e.g., JPG, PNG).
- Video: Supported for object tracking and frame-by-frame inference.
Output
The model outputs bounding boxes, class labels, and confidence scores for each detected object.
Example Results
Here are some example results generated by the model:
| Input Image | Output Image |
|---|---|
 |
 |
Training Details
- Pretrained Weights: YOLO pretrained weights were used as the starting point.
- Hyperparameters: Custom hyperparameters for learning rate, epochs, and batch size were used. Refer to the
config.yamlfile for details. - Hardware: Training was conducted on NVIDIA RTX 3090.
Demo
You can interact with this model through a hosted Gradio app:
Gradio Demo (if available).
Limitations
- The model is optimized for the specific dataset it was trained on. It may not generalize well to other datasets.
- Objects not present in the training dataset may not be detected accurately.
License
This model is licensed under the MIT License. Feel free to use it for personal or commercial purposes with attribution.
Acknowledgments
- Built using Ultralytics YOLO.
- The dataset used for model training is taken from public datasets such as public streams or clips without including any personal information.
Contact
For questions, issues, or suggestions, feel free to reach out:
- Author: Frendy Rachman (https://huggingface.co/frendyrachman)
- Email: [email protected]