title = {PyCinemetricsV2: Interactive computational film software based on transformers and PySide6}, journal = {SoftwareX}, volume = {31}, pages = {102299}, year = {2025}, issn = {2352-7110}, doi = {https://doi.org/10.1016/j.softx.2025.102299}, url = {https://www.sciencedirect.com/science/article/pii/S2352711025002651}, author = {Chunfang Li and Yalv Fan and Yushi Shen and Kun Wang and Yuhe Hu and Fei Zhang and Yuchen Pei and Tongtong Zheng and Zhuoqi Shi}
- After cloning, the only missing folder is
models. After downloading themodelsfolder from the cloud drive, just place it in the directory, and it will run. - Files shared through the cloud drive:
Model files and test_videos:
- Link: Baidu Netdisk, Google Drive
- Model and functionality correspondences:
- Video boundary detection: transnetv2
- Face recognition: buffalo_l
- Speech-to-text subtitles: faster-whisper-base
- Subtitle detection:paddleocr
- Object detection: git-base
- Translation: opus-mt-en-zh
- Shot type recognition:pose net
- Potential issues:
- If
pip install insightfacefails, please install it manually. Link to GitHub - Some models are outdated, so the latest numpy cannot be used. Version 1.26.0 is compatible. 5.License:
- This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License (LGPL) as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version.
- If
- clone之后仅缺少
models文件夹,网盘下载models文件夹后放进目录即可运行。 - 通过网盘分享的文件: 模型文件和测试视频:
- 模型与功能对应
- 视频边界检测:transnetv2
- 人脸识别:buffalo_l
- 语音识别字幕:faster-whisper-base
- 字幕检测:paddleocr
- 目标检测:git-base
- 翻译: opus-mt-en-zh
- 镜头类型识别:pose net
- 可能会遇到的问题
- 如果pip install insightface出错,请手动安装。 Link to GitHub
- 因为有些模型比较旧,所以不能使用最新的numpy,1.26.0是可用的。
软件
https://pan.baidu.com/s/1_4PZGLwli_wLjY3xEPO6uQ?pwd=2025 提取码: 2025
