Table of Contents
Auto-tagging
Learn how auto-tagging works and explore the available features.
Table of Contents
Auto-tagging
In Fotoware, auto-tagging refers to the automated process of assigning descriptive tags to images.
Using Artificial Intelligence (AI), Fotoware can automatically detect keywords, brands, and faces and extract text content from images. Fotoware uses Microsoft Azure AI services for AI-powered image analysis. For more information, see Cognitive Services—APIs for AI Solutions | Microsoft Azure. Auto-tagging uses image analysis features provided by Azure Vision.
Fotoware sends images to Azure Vision by using either an action or an asset webhook. The tagging results are saved to selected metadata fields on the asset in Fotoware. Auto-tagging is only available for image files; the file sent to Azure Vision is always a JPEG.
In Fotoware, auto-tagging is configured as an action or asset-ingested webhook. Both use the same URL format to trigger auto-tagging, and the configuration settings are included in the URL in both cases. For more information, see Configuring auto-tagging. Since auto-tagging results are machine-generated, we recommend that they be reviewed by a person and tagged on demand.
Fotoware supports Azure Vision Image Analysis 4.0.
Data and privacy
Microsoft automatically deletes your images and videos after processing and does not use your data to train the underlying models. Video data does not leave your premises, and video data is not stored on the edge where the container runs. For more information, see Azure AI Vision with OCR and AI | Microsoft Azure (FAQ section).
Note
Auto-tagging is a billable feature. For more information, contact Fotoware.
Features
Customers can configure which features they would like to use and mix and match features as necessary.
The table lists the available features and their corresponding functions.
| Feature | Description |
|---|---|
|
Description |
Auto-tagging can analyze an image and generate a human-readable description of its content. The algorithm returns several descriptions based on different visual features, each with a confidence score. The final output is a list of descriptions ordered from highest to lowest confidence. Azure Vision only returns descriptions in English, but we use Azure Translator services to translate them into all languages. Image Analysis can return content tags for thousands of recognizable objects, living beings, scenery, and actions that appear in images. Tagging is not limited to the main subject, such as a person in the foreground; it also includes the setting (indoor or outdoor), furniture, tools, plants, animals, accessories, gadgets, and so on. |
|
Tags |
Image Analysis can return content tags for thousands of recognizable objects, living beings, scenery, and actions that appear in images. Tagging is not limited to the main subject, such as a person in the foreground; it also includes the setting (indoor or outdoor), furniture, tools, plants, animals, accessories, gadgets, and so on. |
|
Optical Character Recognition (OCR) |
OCR is also referred to as text recognition or text extraction. With machine-learning-based OCR techniques, you can extract printed or handwritten text from images, such as posters, street signs, and product labels. The text is typically extracted as words, text lines, paragraphs, or a given image contains brand logos; if so, it enables access to the digital version of the scanned text. This eliminates or significantly reduces the need for manual data entry. |
|
Brands/Logos |
Brand detection uses a database of thousands of global logos to identify commercial brands in images. The Computer Vision service detects whether a given image contains brand logos; if so, it returns the brand name. Note that this feature identifies company logos in images. Brand names, for instance, will not be identified but could be picked up by OCR. |
|
Faces |
Image Analysis can detect if there are human faces within an image and return a true/false value. |
|
People |
Image Analysis can detect people appearing in images. The feature returns the number of people detected, or the text No people detected if none are detected. |
|
Objects |
Object detection is similar to tagging, but the feature returns a tag for each object found in the image. For example, if an image contains a dog, a cat, and a person, the object detection operation will list each object. |
Languages supported
Azure Vision supports different languages depending on the Image Analysis feature. For more information, see What is Image Analysis?.
To support the same range of languages across all auto-tagging features in Fotoware, we use Azure Translator to translate text from English into languages not natively supported by Azure Vision. For information about the languages supported by Azure Translator, see Azure Translator language support.