Add screenparser UI-element detection predictions

#2
by Burhan-Q - opened
Voxel51 org

Adds a screenparser_preds field containing UI-element detection predictions (93,752 detections across 739 images) generated by the screenparser remote FiftyOne zoo model.

Source model:

Each prediction includes label, bounding_box, and confidence. Original ground-truth fields are unchanged; only the new screenparser_preds field is added. Loadable via fiftyone.utils.huggingface.load_from_hub.

harpreetsahota changed pull request status to merged

Sign up or log in to comment