equinor/label-ai

AI Summary: Implement a visual confirmation message (e.g., a brief pop-up or tooltip) to indicate successful copying to the clipboard after a user clicks the 'LABEL IT' button in the label-ai image labeling application. The message should be easily noticeable but not disruptive.

AI Summary: Enhance the existing image and text labeling application to support multiple simultaneous tags or transformations for text. Currently, only one tag/transformation can be applied; the update should allow combining multiple, while acknowledging potential mutual exclusivity between certain transformations (e.g., LTR/RTL tagging of words and sentences).

AI Summary: Enhance the image labeling app to support additional text replacement options beyond the current implementation. This includes handling dashes, tags (potentially from the Private Use Area), and homoglyphs, while acknowledging the fragility of homoglyph substitution. The goal is to expand the app's capabilities for invisible text labeling.

AI Summary: Implement functionality in the 'label-ai' application to allow users to add Content Credentials (CC) metadata to images, enabling the creation and embedding of CC manifests. This involves integrating a suitable Python library (like the C2PA Python library) and potentially handling certificate management for Equinor's trusted authorities. The goal is to extend the app's capabilities beyond just reading existing CC data to also writing it for original images.

AI Summary: Implement support for Content Credentials (CR) in the label-ai application. This involves reading CR data from existing images (if present, particularly ChatGPT-generated images) and, optionally, writing CR data to original images. The challenge lies in determining the feasibility and implications of writing CR data for all Equinor images.

AI Summary: Implement functionality to add source and prompt information to the EXIF data of images saved via right-click > save image as, leveraging existing free text fields if possible. This ensures metadata about the AI-generated image (source and prompt used) is preserved alongside the image itself.