app

Extract frames from videos

Extract Frames from Videos

OverviewHow To Use

GitHub release (latest SemVer) views used by teams runs

Overview

App allows to extract video frames to images project without labels. User should define frames step, and datasets structure (keep original structure or save frames for every video to separate dataset) and result project name.

Images will have name in the following structure: {original video id}_frame_{frame index}.jpg. Also additional meta information is attached to every frame and available in image labeling tool. For example:

{
    "original_project_id": 333,
    "original_project_name": "my-video-project",
    "original_dataset_id": 444,
    "original_dataset_name": "roads",
    "original_video_id": "7484736",
    "original_video_name": "cars.mp4",
    "frame": 15
}

How To Use

Step 1: Add app to your team from Ecosystem if it is not there

Step 2: Run app from context menu of video project or dataset:

Step 3: Define settings

Step 4: Wait until task is finished. Result project is created in the same workspace.

Every extracted frame has metainformation about original video:

Repository

Last Updated 2020-12-30 21:14:38
Author mkolomeychenko