supervisely-ecosystem / turn-video-project-into-images

Converts labeled video project to images

Videos project to images project

Overview • How To Use • Results

Overview

Transforms supervisely videos project to supervisely images project.

Application key points:

• Backward compatible with Images project to videos project
• Tag named object_id is used for backward compatibility with Images project to videos project
• Dataset structure and names remain unchanged
• Result project name format: "{original_project_name}(images)"
• Image name format: "{video_name}_frame_{:05d}.jpg", for example "my_video_frame_00077.jpg"
• Video tags (both properties and frame range tags) are assigned to corresponding images and objects
• Information about original video project (video_id, video_name, frame_index, video_dataset_id, video_dataset_name, video_project_id, video_project_name) is assigned to every image as metadata.

How To Use

1. Add Videos project to images project to your team from Ecosystem.

2. Run app from the context menu of Videos Project:

3. Select whether you need all or only annotated video frames, select batch size, datasets and set frame step if needed. Press the Run button to launch.

Results

After running the application, you will be redirected to the Tasks page.
Once application processing has finished, your project will be available.
Click on the project name to proceed to it.

turn-video-project-into-images's Issues

Process videos without annotations

api.video.download_path(video_info.id, video_path) - add progress