์ ์ฒด | ์๋น์ค ์๋๋ฆฌ์ค ์ ์ ๋ฐ ๊ตฌ์ฒดํ |
---|---|
๊ตฌ์๋ชจ | oemer custom, Web Back-end(์ฌ์ฉ์ ์ธ์ฆ) |
๊ถ๊ท๋ณด | Model pipeline, Model training, convert to XML |
๋ฐฐ์ํ | dataset research, Web Front-end |
์คํ๋ณ | model pre-processing, model post-processingโ |
์ด์์ง | Web Back-end(DB, server) |
# clone repo
git clone https://github.com/boostcampaitech4lv23cv1/level3_productserving-level3-cv-02.git
# install requirements
pip install -r requirements.txt
# install to convert mp3 package
cd MusicXML2Audio/
sh install_packages.sh
# Run
cd ../app/
python main.py
https://tuggeluk.github.io/deepscores/
# Example code
.
|-- deepscores_test.json
|-- deepscores_train.json
|-- images
| |-- image1.png
| |-- image2.png
| `-- ...
|-- instance
| |-- image1.png
| |-- image2.png
| `-- ...
|-- segmentation
| |-- image1.png
| |-- image2.png
| `-- ...
|-- deepscores_test.json
`-- deepscores_train.json
https://github.com/WongKinYiu/yolov7
dataset์ ๊ทธ๋๋ก ํ์ต์ํฌ ๊ฒฝ์ฐ ๋ถํ์ํ ์ ๋ณด๊ฐ ๋ง์ ์ ์ฒ๋ฆฌ ๊ณผ์ ์ด ํ์ํ์ต๋๋ค.
์ ์๊ธฐํธ์ ์ํ๋ง ๋จ๊ธฐ๊ธฐ ์ํด ์ค์ ์ง์ ์์น ์ ๋ณด๋ง ์ ์ฅํ ํ ์ญ์ ํฉ๋๋ค. ์ดํ ์ ๋ณด๋ฅผ ์ด์งํํ์ฌ ํ์ํ ์ ๋ณด๋ง ์ถ์ถํ ํ ํ์ต์ํต๋๋ค.
model์ prediction์ ๊ธฐ๋ฐ์ผ๋ก ์ฐ์ฃผํ ๋ ํ์ํ ์ ๋ณด๋ฅผ ๊ณ์ฐํฉ๋๋ค.
- conversion from yolo: ์ ๊ทํ๋ bbox ์ขํ๋ฅผ ์ผ๋ฐ ์ ๋ณด ์ขํ๋ก ๋ณํํฉ๋๋ค.
- merge bbox: ํ ์ํ๋ ๊ธฐํธ๋ฅผ ์ฌ๋ฌ๋ฒ ์์ธกํ๋ค๋ฉด ํ๋๋ก ํฉ์นฉ๋๋ค.
- noise removal: ์์ธก ๊ฒฐ๊ณผ์ noise๋ฅผ ์ ๊ฑฐํฉ๋๋ค.
- beat detection: ์ํ์ ๋ฐ์ ์ ๋ณด๋ฅผ ๊ณ์ฐํฉ๋๋ค.
- measure calculation: ์ํ๊ฐ ๋ช๋ฒ์งธ ๋ง๋์ ์ํด์๋์ง ๊ณ์ฐํฉ๋๋ค.
- pitch detection: ๊ณ์ด๋ฆ์ ๋ฐํํฉ๋๋ค.
- SFN(Sharp Flat Natural) detection: ์กฐํ๋ฅผ ์ ์ฉํฉ๋๋ค.
post processing ๊ฒฐ๊ณผ๋ฅผ MusicXML ํ์์ผ๋ก ๋ณํํฉ๋๋ค.
app/
.
|-- constant.py
|-- db
| |-- __init__.py
| |-- connection.py
| |-- core
| | |-- __init__.py
| | |-- config.py
| | |-- db_login.py
| | `-- key.json
| |-- crud
| | |-- __init__.py
| | |-- image_bundle.py
| | |-- sound.py
| | `-- users.py
| |-- models
| | |-- __init__.py
| | |-- image.py
| | |-- image_bundle.py
| | |-- sound.py
| | `-- users.py
| |-- routes
| | |-- __init__.py
| | |-- image_bundle.py
| | |-- sound.py
| | `-- users.py
| |-- schemas
| | |-- __init__.py
| | |-- image.py
| | |-- image_bundle.py
| | |-- sound.py
| | `-- users.py
| |-- service
| | |-- __init__.py
| | |-- image_bundle.py
| | |-- sound.py
| | `-- users.py
| `-- session.py
|-- main.py
|-- output
|-- poetry.lock
|-- pyproject.toml
|-- secret.py
|-- service.py
|-- static
| |-- css
| | |-- css files
| `-- ******.css
| |-- fonts
| `-- font files
| `-- js
| | |-- js files
| `-- ******.js
|-- templates
| |-- html files
| `-- ******.html
`-- utils.py
MusicXML2Audio/
.
|-- constant.py
|-- converter
| |-- MXL2midi.py
| |-- midi2wav.py
| `-- wav2sound.py
|-- data
|-- install_packages.sh
|-- main.py
`-- midi2audio.py
- DeepScores -- A Dataset for Segmentation, Detection and Classification of Tiny Objects https://arxiv.org/pdf/1804.00525.pdf
- Understanding Optical Music Recognition https://arxiv.org/abs/1908.03608
- https://github.com/yvan674/obb_anns
- https://github.com/BreezeWhite/oemer/tree/main/oemer
- https://github.com/FluidSynth/fluidsynth
- https://github.com/bzamecnik/midi2audio