Video Dataset Creation

Video dataset creation in Pro involves extracting frames from videos for frame-level annotation and AI model training on tasks such as detection and tracking. The platform also provides duplicate frame removal and image conversion features to ensure high-quality training data

💡Steps to access and manage datasets

• Follow the same steps outlined in “Section 2.1: Vision Dataset Creation” up to the dataset upload stage. At the upload step, select ‘Video’ instead of ‘Data’

• After uploading, select the ‘Frames’ option, set the desired frame range, and click on ‘Extract Frames’

• After uploading, a toast message will appear upon completion of frame extraction

  • Click on the ‘Remove Duplication’ tag and wait for the upload to complete.

  • Wait a moment while the upload is processed. Once complete, a pop-up message will appear indicating ‘Completed’

  • Now it will direct , A summary card is shown the, displaying:

    • Total Duplicates Removed – Number of duplicate images eliminated.
    • Total Unique Images Remaining – Number of images that passed the duplicate check and are available for further use.

  • Extracted video frame images will now be available in the dataset.