DeepCut
DeepCut is a collection of Python scripts designed for automated video processing tasks such as extracting frames, detecting actions, removing captions, and more. It aims to streamline common media-processing tasks into a single, flexible pipeline.
Table of Contents
Features Getting Started Scripts Overview Usage Dependencies Contributing License Features
Frame Extraction: Quickly extract frames from video files. Action Detection: Automatically detect specific actions in video footage. Caption Removal: Remove subtitles/captions from video frames or video files. Silence Removal: Detect and remove silent sections of video/audio. Web Crawler: (If applicable) Crawl websites or data sources for relevant media files. Getting Started
Clone the repository: git clone git@github.com:Xavierhuang/DeepCut.git Navigate to the project folder: cd DeepCut Install any required dependencies (see Dependencies). Scripts Overview
Script Description 1_extract_frames.py Extracts frames from a video at specified intervals. action_detector.py Detects certain actions or events in a video. remove_captions.py Removes or masks captions/subtitles within a video. remove_silence.py Detects and removes silent portions in audio tracks. crawler.py (If used for data collection) Crawls sources to fetch or download files. 1_extract_frames.py Extract frames from your video:
python 1_extract_frames.py --input /path/to/video.mp4 --output /path/to/frames --interval 10 --input: Path to the input video file --output: Folder where extracted frames will be saved --interval: Number of frames or seconds between extractions (depending on implementation) action_detector.py Detect specified actions:
python action_detector.py --video /path/to/video.mp4 --model /path/to/model --video: Path to the video --model: Path to your trained action-detection model remove_captions.py Remove or blur out captions:
python remove_captions.py --video /path/to/video.mp4 --output /path/to/processed_video.mp4 remove_silence.py Strip silent sections from a video or audio track:
python remove_silence.py --input /path/to/video.mp4 --threshold 0.01 --threshold: Decibel or amplitude threshold for detecting silence crawler.py Example usage:
python crawler.py --url https://somesite.com/videos --output /path/to/data Usage
Choose a script you want to run. Review script arguments (--help often lists these): python 1_extract_frames.py --help Run the script with the desired parameters. Dependencies
Python 3.x (e.g., 3.7+) Required libraries: (Example below—adjust to your actual needs) opencv-python numpy requests moviepy etc. Install via pip:
pip install -r requirements.txt (Create a requirements.txt listing your exact dependencies if you haven’t already.)
Contributing
Contributions are welcome!
Fork the project. Create a feature branch. Commit your changes. Push to your fork. Create a new Pull Request. License
You can include a license of your choice here (e.g., MIT, Apache 2.0). For example:
MIT License
Copyright (c) 2025
Permission is hereby granted... [Full text of the license] Feel free to modify this structure as needed for your specific project. If you have unique features, custom usage steps, or want to detail your dataset, be sure to add those sections.