Plugin errors on execution; superseded by built-in Auto Labeling in FiftyOne

[mgustineli]

Summary

The zero-shot prediction plugin errors out during operator execution on FiftyOne Enterprise (v2.14.1). Additionally, this plugin's core functionality is now superseded by FiftyOne's built-in Auto Labeling feature, which ships natively with recent FiftyOne releases.

Error

When attempting to run the plugin, the following error occurs in the browser console:

▼ Error occurred during operator execution

83016/en/e/<@https://<deployment>.fiftyone.ai/_next/static/chunks/pages/_app-bd30142cbbbafa19.js:29:393497
h@https://<deployment>.fiftyone.ai/_next/static/chunks/main-672e8d30757c054a.js:1:52830
22037/o/d/<.value<@https://<deployment>.fiftyone.ai/_next/static/chunks/main-672e8d30757c054a.js:1:52609
22037/o/S/</<@https://<deployment>.fiftyone.ai/_next/static/chunks/main-672e8d30757c054a.js:1:53236
r@https://<deployment>.fiftyone.ai/_next/static/chunks/pages/_app-bd30142cbbbafa19.js:27:2077222
s@https://<deployment>.fiftyone.ai/_next/static/chunks/pages/_app-bd30142cbbbafa19.js:27:2077425

Environment

• FiftyOne Enterprise: v2.14.1
• Plugin: installed via fiftyone plugins download

Built-in Alternative: Auto Labeling

FiftyOne now ships a native Auto Labeling panel (currently in Beta) that covers the same use cases as this plugin and more:

• Classification, Detection, and Instance Segmentation task types
• Zero-shot and Fixed vocabulary model types
• Built-in support for CLIP, SigLIP, and other models from the FiftyOne Model Zoo
• Configurable target scope (full dataset, current view, or selection)
• Settings for label field, minimum confidence, run naming, and batch processing

Since Auto Labeling is now part of FiftyOne's core product, this plugin may be a candidate for deprecation or an archival notice in the README pointing users to the built-in feature.

Suggestion

Consider one of the following:

1. Archive the repo with a note in the README redirecting users to FiftyOne's built-in Auto Labeling panel
2. Add a deprecation notice to the README explaining that this functionality is now available natively
3. Fix the runtime error if there's still a reason to maintain the plugin alongside the built-in feature