11# Evaluation Framework for CCMusic Database Classification Tasks
2- [ ![ License: MIT] ( https://img.shields.io/github/license/monetjoe /ccmusic_eval.svg )] ( https://github.com/monetjoe /ccmusic_eval/blob/main/LICENSE )
3- [ ![ Python application] ( https://github.com/monetjoe /ccmusic_eval/actions/workflows/python-app.yml/badge.svg?branch=main )] ( https://github.com/monetjoe /ccmusic_eval/actions/workflows/python-app.yml )
2+ [ ![ License: MIT] ( https://img.shields.io/github/license/ccmusic-database /ccmusic_eval.svg )] ( https://github.com/ccmusic-database /ccmusic_eval/blob/main/LICENSE )
3+ [ ![ Python application] ( https://github.com/ccmusic-database /ccmusic_eval/actions/workflows/python-app.yml/badge.svg?branch=main )] ( https://github.com/ccmusic-database /ccmusic_eval/actions/workflows/python-app.yml )
44[ ![ hf] ( https://img.shields.io/badge/HuggingFace-ccmusic-ffd21e.svg )] ( https://huggingface.co/ccmusic-database )
55[ ![ ms] ( https://img.shields.io/badge/ModelScope-ccmusic-624aff.svg )] ( https://www.modelscope.cn/organization/ccmusic-database )
66[ ![ arxiv] ( https://img.shields.io/badge/arXiv-2503.18802-b31b1b.svg )] ( https://arxiv.org/pdf/2503.18802.pdf )
@@ -12,7 +12,7 @@ Classify spectrograms by fine-tuned pre-trained CNN models.
1212
1313## Download
1414``` bash
15- git clone git@github.com:monetjoe /ccmusic_eval.git
15+ git clone git@github.com:ccmusic-database /ccmusic_eval.git
1616cd ccmusic_eval
1717```
1818
@@ -34,7 +34,7 @@ python train.py --ds ccmusic-database/bel_canto --subset eval --data cqt --label
3434| --subset | The subset of the dataset | For examples: default, eval | string |
3535| --data | Input data colum of the dataset | For examples: mel, cqt, chroma | string |
3636| --label | Label colum of the dataset | For examples: label, singing_method, gender | string |
37- | --model | Select a [ CV backbone] ( https://huggingface.co/datasets/monetjoe /cv_backbones ) to train | [ Supported backbones] ( https://www.modelscope.cn/datasets/monetjoe /cv_backbones/dataPeview ) | string |
37+ | --model | Select a [ CV backbone] ( https://huggingface.co/datasets/ccmusic-database /cv_backbones ) to train | [ Supported backbones] ( https://www.modelscope.cn/datasets/ccmusic-database /cv_backbones/dataPeview ) | string |
3838| --imgnet | ImageNet version the backbone was pretrained on | v1, v2 | string |
3939| --mode | Training mode ID | 0=linear_probe, 1=full_finetune, 2=no_pretrain | int |
4040| --bsz | Batch size | For examples: 1, 2, 4, 8, 16, 32, 64, 128..., default is 4 | int |
0 commit comments