Export data

Exporting your labeled data for advanced analysis and ML training

To export a CSV or JSON file of your Image, Object, and Label metadata, perform the following steps:

1. Select the View/Filters you'd like to apply to your export

2. Click on the "Export Data" button on the bottom of the Filters panel (bottom left of screen)

3. Select which format you'd like to export your data in

Only reviewed images will be exported

Currently, even if the "non-reviewed" filter is checked on, only fully reviewed Images and Objects will be exported. Similarly, only the "winning" Labels on Objects (the most recently validated Label) will be exported. All invalidated or non-validated Labels are ignored.

Timezones

Be mindful that all date-time fields in all exports will be in UTC+0.

Supported export formats

CSV

CSV export is helpful for ecological data analysis in Excel and CamtrapR. It does not contain bounding-box information for the Objects.

COCO for Camera Traps

COCO for Camera Traps is helpful for pulling Image and annotation data into a machine learning training workflow. The Object records (annotations) do contain bounding box information.

Exporting image files

Currently you need to have special AWS permissions and some basic familiarity with a command line interface to export/download your image files. If you need access, please reach out to Nathaniel at [e-mail] for support.

The image export process entails the following steps:

1. Procure the necessary AWS credentials

2. Install the AWS CLI and configure a new named "profile" (the profile name must be animl)

3. Clone and follow the setup instructions in the animl-analytics GitHub repository

4. Run the utils/download_images.py script as described in the animl-analytics documentation