Image Labelling is an AI-powered feature in the Egnyte Mobile App that automatically generates a short description and object labels for images uploaded from the device. Analysis is performed entirely on-device using a locally installed AI model — the image is not sent to any external server.
The generated caption and labels are stored as file metadata in Egnyte and fed into the search index, making images discoverable through natural language queries. This is particularly useful for teams in construction and engineering who capture large volumes of job site photos that would otherwise be difficult to organize or retrieve.
The feature is currently available in the Android application (and expected to be added to the iOS app at the beginning of October 2026).
Availability
- Platforms: Android
- Minimum app version: Android 9.11 or higher
- Plan availability: AEC Essentials, AEC Elite, AEC Ultimate
- Processing: On-device (no data sent to external AI services)
How it Works
The AI Image Labelling toggle is available in the Egnyte Mobile App under Settings → AI Features → AI Image Labelling. When the toggle is enabled, every image uploaded through the app is automatically analyzed on the device.
The AI model identifies objects, materials, equipment, and site conditions in the photo, and saves a concise description along with individual object labels as file metadata in Egnyte. The description is then indexed, making the image searchable by content.
When a user on a supported domain upgrades to version 9.11 or higher for the first time, the app prompts them to download the on-device AI model.
The model can also be downloaded at any time through the settings path above.
Supported File Types and Behavior
Image Labelling processes standard image files — including JPEG, PNG, HEIC, and HEIF — uploaded through the Egnyte Mobile App. Analysis runs at upload time, on the device. The generated caption is saved as file metadata and is visible in the file details screen in the Tags category.
Image Labelling applies only to images uploaded after the toggle is enabled. Previously uploaded images are not retroactively processed.
Searching by Image Content
Once images have been labelled, their descriptions are included in Egnyte's search index. Searches for terms such as “caterpillar”, "excavator," or "rebar" will surface matching images alongside other file types — no manual tagging required.