Guides
Vi Dashboard
  1. Datasets
  2. Manage Datasets
  3. Download Data

Download Full Dataset

Export your complete dataset including all assets and annotations from Datature Vi for backup, local development, or data migration.

Prepare Data
Train a Model
Deploy & Integrate
Manage Workspace

Downloading the full dataset exports all assets (images and videos) together with their annotations in a structured folder. This is the right choice when you need a complete backup, want to train a model locally, or are migrating data to another platform.

Before You Start
  • You need an existing dataset in your Datature Vi workspace.
  • Make sure you have enough local disk space. A dataset with 1,000 images typically needs 500 MB to 5 GB depending on resolution.
  • For datasets over 1,000 assets, the Vi SDK provides more reliable downloads with automatic retry logic.
1

Go to the Annotations tab

Go to the Annotations tab

Click the Explorer tab in the sidebar, select your dataset, then click the Annotations tab in the Dataset Explorer header. The page displays the Export Annotations section with an Export button.

You should see
Dataset export page showing a finished job

Your dataset is exported when you see a successfully finished job. You can then download the dataset.

Exported folder structure

With train/test split enabled (ratio > 0):

  • dataset-name
    • train
      • images
        • image1.jpg
        • image2.jpg
      • annotations
        • image1.jsonl
        • image2.jsonl
    • test
      • images
        • image3.jpg
      • annotations
        • image3.json
    • metadata.json

Without split (ratio = 0 or disabled):

  • dataset-name
    • dump
      • assets
        • image1.jpg
        • image2.jpg
      • annotations
        • image1.jsonl
        • image2.jsonl
    • metadata.json

Track your export in job history

The Annotation Job History section shows all export operations:

  • Job type: Upload or Export
  • Initiated by: Who started the export
  • Status: In Progress, Finished, or Failed

Troubleshooting

  • Datasets with thousands of assets can take several minutes to process. Check the Annotation Job History to confirm the job is still running.
  • For large datasets, use the Vi SDK which handles large downloads more reliably.
  • Try downloading during off-peak hours if server load is high.
  • Use the job history to re-download the completed export instead of retrying from scratch.
  • Check that you have sufficient disk space before downloading.
  • Try a different browser or clear the browser cache, then re-download.
  • Confirm no filters are active in the Dataset Explorer that might be limiting the export scope.
  • Verify the asset and annotation counts in the dataset overview match your expectations before exporting.
  • Re-export if the counts look wrong after the first attempt.
  • Unzip the downloaded file if it arrived in compressed format.
  • Use a UTF-8 compatible text editor or JSON tool to open annotation files.
  • Check that you have read permissions on the extracted files.

Do this with the Vi SDK

import vi

client = vi.Client(
    secret_key="your-secret-key",
    organization_id="your-organization-id"
)

result = client.datasets.download(
    dataset_id="your-dataset-id",
    save_dir="./data"
)
print(result.summary())

For more details, see the full SDK reference.

Next steps

Download Annotations Only

Export annotation files without asset media for a lightweight backup.

Vi SDK Getting Started

Download, load, and iterate through datasets programmatically.

Train A Model

Use your downloaded dataset to train a model locally or on Vi.

Updated about 1 month ago