https://github.com/mbari-org/deepsea-frameextractor

FFMPEG based docker image for extracting still images from video from timecodes, strides, etc. for labeling tasks. https://github.org/mbari-org/deepseat-frameextractor
https://github.com/mbari-org/deepsea-frameextractor

ffmpeg underwater-images underwater-robotics

Last synced: about 1 month ago
JSON representation

FFMPEG based docker image for extracting still images from video from timecodes, strides, etc. for labeling tasks. https://github.org/mbari-org/deepseat-frameextractor

• Host: GitHub
• URL: https://github.com/mbari-org/deepsea-frameextractor
• Owner: mbari-org
• License: apache-2.0
• Created: 2020-06-26T00:00:25.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2022-07-06T20:46:39.000Z (over 2 years ago)
• Last Synced: 2023-03-04T16:39:30.914Z (almost 2 years ago)
• Topics: ffmpeg, underwater-images, underwater-robotics
• Language: Python
• Homepage:
• Size: 22.5 KB
• Stars: 1
• Watchers: 4
• Forks: 1
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • Changelog: CHANGELOG.md
  • License: LICENSE

Awesome Lists containing this project

README

        
[![MBARI](https://www.mbari.org/wp-content/uploads/2014/11/logo-mbari-3b.png)](http://www.mbari.org)

[![semantic-release](https://img.shields.io/badge/%20%20%F0%9F%93%A6%F0%9F%9A%80-semantic--release-e10079.svg)](https://github.com/semantic-release/semantic-release)

![Supported Platforms](https://img.shields.io/badge/Supported%20Platforms-Windows%20%7C%20macOS%20%7C%20Linux-green)

![license-GPL](https://img.shields.io/badge/license-GPL-blue)

# About

Frame extractor to convert movies into still images for labeling.  

Extracts still frame per duration specified in --step of a movie, (optionally) deinterlaces, and saves as a png.

Uses multiple cores to speed-up computation.

# Questions?

If you have any questions or are interested in contributing, please contact me at [email protected].

*Danelle Cline*

https://www.mbari.org/cline-danelle-e/

---

  

## Run Examples

*Arguments*

  * -i input directory of video files(.mov, .mp4, anything ffmpeg will support), or single input video files; if specifying input directory must use keys

  * -o (optional, defaults to input directory) output directory to store results. 

    * Two directories are created for the results:

     * the key name, or the root filename, e.g. D0232_20160501T000030Z if input is D0232_20160501T000030Z.mov

     * imgs

  * -t or --start Start time in HH:MM:SS format (optional)

  * -e or --end End time in HH:MM:SS format (optional)

  * -m or --milliseconds Number of milliseconds to capture every step

  * -s or --step Step size in seconds to grab a frame. If not specified, extracts all frames

  * -g glob pattern list of search patterns to find video files in conjunction with -i

  * -d (optional) deinterlace method, e.g. drop or yadif

    * drop will remove every other field

    * yadif = Yet Another Deinterlacing Filer (see ffmpeg page for details on this)

    

*Examples*

Extract one frames every 5 seconds from file D0232_20160501T000030Z.mov to your desktop, no deinterlacing

```bash

docker run -v $PWD/data:/data  -v /Users/dcline/Desktop:/desktop mbari/deepsea-frameextractor -i /data/D0232_20160501T000030Z.mov -o /desktop

```

Extract one frame every 5 seconds from file D0232_20160501T000030Z.mov to your desktop, drop deinterlacing

```bash

docker run -v $PWD/data:/data  -v /Users/dcline/Desktop:/desktop mbari/deepsea-frameextractor -i /data/D0232_20160501T000030Z.mov -o /desktop  -d drop

```

Extract one frame every 2 seconds between 00:00:00 to 00:05:00 of file D0232_20160501T000030Z.mov to your desktop

```bash

docker run -v $PWD/data:/data  -v /Users/dcline/Desktop:/desktop mbari/deepsea-frameextractor -i /data/D0232_20160501T000030Z.mov -o /desktop -s 2 --start 00:00:00 --end 00:05:00

```

Extract 2 seconds of each video to your desktop, searching for videos matching the pattern D*.MOV recursively in the /data directory. 

This will split compute for each video across each CPU available.

```bash

docker run -v $PWD/data:/data  -v /Users/dcline/Desktop:/desktop mbari/deepsea-frameextractor -i /data  --keys '/**/D*.MOV' -o /desktop -s 2

```

## Developer notes 

Run interactively, mounts your current directory in the container as tmp/code, and any

changes made in that /tmp/code directory, even after the container closes will persist.

```bash

docker run -v $PWD:/tmp/code -it --entrypoint='bash' mbari/deepsea-frameextractor

```

### Exiftools

- All tags available by format: https://sno.phy.queensu.ca/~phil/exiftool/TagNames/index.html

- For PNG https://sno.phy.queensu.ca/~phil/exiftool/TagNames/PNG.html