How to export AI Training Data

This how-to explains how to generate an AI Training Data using the Tygron Platform. This data can be used to train your own AI model.

How to export AI Training Data:

1. Open a project in editor mode using the Tygron Client Application
2. Optionally add a Satellite Overlay. This overlay will be used to export train images for the dataset.
3. Determine what [item] type will be used to export a dataset for. For example Foliage areas drawn on top of the satellite overlay.
4. Select the Export Geo Data option in the Ribbon bar popup of the selected item type. For example: Hover over Areas and select Export Geo Data
5. Under Format, select the option AI Training Data PNG
6. Filter items with an attribute, such as FOLIAGE
7. Optionally select a prepared Intersection Area, for example a TRAIN_LIMIT_AREA.
8. For the Overlay, select the preferred Overlay, for example the Satellite Overlay.
9. The image size can increased, as long as the amount of expected features within the image does not exceed the maximum amount of 250.
10. The stride can be kept at 50%
11. Click on the Export Files button
12. Select a suitable folder for the generated dataset.
13. Wait until the dataset is fully generated.

• 
  Configuration of the Export AI Training Data as PNG's for Foliage Areas on a Satellite Overlay, with an intersecting area.

Notes

• In case the export folder is not empty, a confirmation popup will appear on overriding the existing image, label, mask and bounding box files.
• In case a region intersects the limit area, but does not contain any features, it is still exported as a dataset-entry (image-, label-, mask- and bbox-file) but without labels, mask pixels or bounding box entries.
• In case the feature density is low, the image size can be increased.
• In case the feature sizes are small compared to the image size, the stride can be increased.
• A dataset can also be generated as AI Training Data CSV, which allows users to directly train on float matrices instead of color images.
• When the intended background image is not the satellite image provided by default but some other image or set of data, it is also possible to select a WMS Overlay referencing an external map service, or a Combo Overlay, the COLOR function of which can be used to compute a custom image with.

How-to's

• How to train your own AI model for an Inference Overlay

See also

• Demo Training Data Project
• PyTorch
• Satellite Overlay