---
title: Guzheng_Tech99
canonical_url: "https://www.modelscope.cn/models/ccmusic-database/Guzheng_Tech99"
md_url: "https://www.modelscope.cn/models/ccmusic-database/Guzheng_Tech99.md"
repository: ccmusic-database/Guzheng_Tech99
chinese_name: "古筝演奏技法逐帧检测模型"
last_updated: 2026-08-05
license: "MIT License"
downloads: 960
stars: 15
---

# Guzheng_Tech99

> Guzheng_Tech99 - ccmusic-database 在 ModelScope 开源的模型。古筝演奏技法逐帧检测模型是一种基于深度学习技术的创新工具，旨在通过对古筝演奏视频的逐帧分析，精准识别和标注演奏过程中的各种技法动作。该模型能够自动检测手指的按弦、拨弦、滑音等细节动作，并将其与对应的技法类别进行匹配，为古筝教学和演奏研究提供可视化和量化的分析支持。通过高精度的帧级检测，该模型不仅能够帮助学习者更好地理解和掌握古筝演奏技巧，还能为专业演奏者提供技术分析和改进的参考依据，推动古筝艺术的传承与发展。

ccmusic-database/Guzheng_Tech99 是 ModelScope 魔搭社区上的机器学习模型，采用 MIT License 许可。

- **Repository**: ccmusic-database/Guzheng_Tech99
- **License**: MIT License
- **Downloads**: 960
- **Stars**: 15
- **Last updated**: 2026-08-05

Source: https://www.modelscope.cn/models/ccmusic-database/Guzheng_Tech99

---

# Intro 简介
For the 99 recordings, silence is first removed, which is done based on the annotation, targeting the parts where there is no technique annotation. Then all recordings are uniformly segmented into fixed-length segments of 3 seconds. After segmentation, clips shorter than 3 seconds are zero padded. This padding approach, unlike circular padding, is adopted specifically for frame-level detection tasks to prevent the introduction of extraneous information. Regarding the dataset split, since the dataset consists of 99 recordings, we split it at the recording level. The data is partitioned into training, validation, and testing subsets in a 79:10:10 ratio, roughly 8:1:1.

对于这 99 首录音，首先去除静音部分，这是基于标注来完成的，针对的是没有技巧标注的部分。然后，所有录音都被统一分割成固定长度为 3 秒的片段。分割后，对于短于 3 秒的音频片段，采用零填充的方式进行处理。这种填充方法与循环填充不同，是专门为帧级别检测任务而采用的，以防止引入额外信息。关于数据集的划分，由于数据集由 99 首录音组成，我们在录音级别上进行划分。数据被划分为训练集、验证集和测试集，比例为 79:10:10，大致为 8:1:1。

## Demo 在线演示 (推理代码)
<https://www.modelscope.cn/studios/ccmusic-database/Guzheng_Tech99>

## Usage 使用
:modelscope-code[]{type="sdk"}

## Maintenance 维护
```bash
GIT_LFS_SKIP_SMUDGE=1 
```

:modelscope-code[]{type="git"}

## Results 训练结果
|     Backbone      |     Mel     |     CQT     |   Chroma    |
| :---------------: | :---------: | :---------: | :---------: |
|     ViT-B-16      |    0.705    |    0.518    |    0.508    |
|      Swin-T       | **_0.849_** | **_0.783_** | **_0.766_** |
|                   |             |             |             |
|       VGG19       | **_0.862_** |    0.799    |    0.665    |
| EfficientNet-V2-L |    0.783    |    0.812    |    0.697    |
|    ConvNeXt-B     |    0.849    | **_0.849_** | **_0.805_** |
|     ResNet101     |    0.638    |    0.830    |    0.707    |
|   SqueezeNet1.1   |    0.831    |    0.814    |    0.780    |
|      Average      |    0.788    |    0.772    |    0.704    |

## Dataset 数据集
<https://www.modelscope.cn/datasets/ccmusic-database/Guzheng_Tech99>

## Mirror 镜像
<https://huggingface.co/ccmusic-database/Guzheng_Tech99>

## Evaluation 校验
<https://github.com/monetjoe/ccmusic_eval/tree/tech99>

## Cite 引用
```bibtex
@article{Zhou-2025,
  author  = {Monan Zhou and Shenyang Xu and Zhaorui Liu and Zhaowen Wang and Feng Yu and Wei Li and Baoqiang Han},
  title   = {CCMusic: An Open and Diverse Database for Chinese Music Information Retrieval Research},
  journal = {Transactions of the International Society for Music Information Retrieval},
  volume  = {8},
  number  = {1},
  pages   = {22--38},
  month   = {Mar},
  year    = {2025},
  url     = {https://doi.org/10.5334/tismir.194},
  doi     = {10.5334/tismir.194}
}
```