https://github.com/kinwaicheuk/ijcnn2021.github.io
source code and additional information for IJCNN2021
https://github.com/kinwaicheuk/ijcnn2021.github.io
Last synced: about 1 month ago
JSON representation
source code and additional information for IJCNN2021
- Host: GitHub
- URL: https://github.com/kinwaicheuk/ijcnn2021.github.io
- Owner: KinWaiCheuk
- Created: 2020-10-15T03:52:15.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2021-03-05T15:25:44.000Z (about 4 years ago)
- Last Synced: 2025-02-05T02:49:13.316Z (3 months ago)
- Language: HTML
- Homepage:
- Size: 44 MB
- Stars: 1
- Watchers: 2
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# IJCNN2021
[Source code](https://github.com/KinWaiCheuk/IJCNN2021.github.io/tree/main/codes) and [supplementary information](https://kinwaicheuk.github.io/IJCNN2021.github.io/) for IJCNN2021.
The supplementary information is available at: https://kinwaicheuk.github.io/IJCNN2021.github.io/
The source code is available under the `codes` folder.
## Training the mode
### Step 1: Preparing Dataset
The MAPS dataset can be downloaded via https://amubox.univ-amu.fr/index.php/s/iNG0xc5Td1Nv4rR.
Download and unzip the dataset into the `codes` folder.
After unzipping all the files inside, the `ENSTDkAm1` and `ENSTDkAm2` folder should be combined as one single folder `ENSTDkAm`.
### Step 2: Training the models
Run the following scripts to train different models:
* `python train_original.py with `
* `python train_fast_local_attent.py with `
* `python train_simple.py with `
The following arguments are available:
* `device`: choose what device to use. Can be `cpu`, `cuda:0` or any device that is available in your PC. Default `cuda:0`.
* `LSTM`: Train the model with or without the LSTM layer. Either `True` or `False`. Default `True`.
* `onset_stack`: Train the model with or without the onset stack. Either `True` or `False`. Default `True`.
* `batch_size`: Setting the batch size. Default `16`.
The following arguments are for `train_fast_local_attent.py` only
* `w_size`: The attention window size. Default: `30`.
* `attention_mode`: Choosing which feature to attend to. Either `onset` or `activation` or `spec`. Default `onset`.
The PyTorch dataset class `MAPS()` inside each script will process and prepare the dataset if you are running it for the first time.
## Using pre-trained models
### Step 1: Downloading weights
The weights can be downloaded [here](https://sutdapac-my.sharepoint.com/:f:/g/personal/kinwai_cheuk_mymail_sutd_edu_sg/Em-RhkuS7S9Oq9iGU25KixcBQ-Ylh-z1miYa9xmrQZ4KYg?e=Y9vl7X)
### Step 2: Generating the results
The bash files contain all the commanands to obtain the results reported in the paper. Run the following bash scripts to get all the results
* `bash get_table1.sh`
* `bash get_figure1.sh`
* `bash get_table2.sh`
The accuracy reports and the midi files will be saved in the `results` folder.