This project focuses on fine-tuning the pre-trained YOLOv10 model for detecting helmet usage in various safety scenarios. The training process utilizes a specific dataset tailored for helmet safety detection, which addresses the problem of ensuring construction workers are wearing helmets.
To use it, access the file helmet_safety_detection_yolov10_colab.ipynb and follow the instructions to fine-tune the pre-trained YOLOv10 model.
After installation, you will save the best model weights, which can be found in the file best.pt.
$ conda create -n <env_name> -y python=3.11
$ conda activate <env_name>
$ git clone https://github.com/THU-MIG/yolov10.git
$ cd yolov10
$ pip install -q -r requirements.txt
$ pip install -e .
$ pip install -q streamlit
$ cd ..
$ streamlit run app.py