Glory-Day / PianistDesktop

Pianist Desktop

1. Usage

• Option.py

# path : PianistDesktop/packages/option.py

class Option:
    def __init__(self):
        self.__dict__['weights'] = 'models/best.pt'
        # Please put in the image you want to convert.
        self.__dict__['source'] = 'models/data/images'
        self.__dict__['img_size'] = 2048
        self.__dict__['conf_thres'] = 0.8
        self.__dict__['iou_thres'] = 0.5
        self.__dict__['device'] = ''
        self.__dict__['view_img'] = False
        self.__dict__['save_txt'] = True
        self.__dict__['save_conf'] = True
        self.__dict__['save_crop'] = True
        self.__dict__['nosave'] = False
        self.__dict__['classes'] = None
        self.__dict__['agnostic_nms'] = False
        self.__dict__['augment'] = False
        self.__dict__['update'] = False
        # Converted results are stored /run/detect/exp{n}
        self.__dict__['project'] = '/runs/detect'
        self.__dict__['name'] = 'exp'
        self.__dict__['exist_ok'] = False
        self.__dict__['line_thickness'] = 1
        self.__dict__['hide_labels'] = False
        self.__dict__['hide_conf'] = False

• Converted files

exp{n} ------ labels ---- text files
        |
        |---- image files

• Note

# first, put the path of converted files

# path : Pianist/utils/score.py
if __name__ == '__main__':
    # Please put the path of the converted images
    image_path = ""
    # Please put the path of the converted labels
    label_path = ""
    ...

# and then, check folder notes

# path : Pianist/result/notes.txt

---

2. Reference

• DeepScore : https://tuggeluk.github.io/
• Pytorch-Yolo v5 : https://github.com/ultralytics/yolov5
• Roboflow : https://roboflow.com/

---

3. Environments

• Hardware Environment

CPU : 48 vCPU
GPU : NVIDIA Tesla A100 x 4
RAM : 340GB memory
Disk : 20GB balanced persistent disk
OS : Ubuntu 18.04 LTS
Pytorch : 1.8.1 version
CUDA: 11.1 version

• Requirements

pip install -r requirements.txt

• Review

---

4. Streamlit

streamlit run main.py

---

5. VR Content Download

• Review

• Google Drive URL : https://drive.google.com/drive/folders/12N9wUYfFP8Lm0WtIN58A-DIdeR_WiPrm?usp=sharing