Overview of image management in the earlier version of Intelligent Media Management-Intelligent Media Management(IMM)-阿里云帮助中心

A medium image project lets you manage image data and perform centralized operations, such as image processing, analysis, search, and management.

Important

This document is no longer maintained. Use the new version of Intelligent Media Management.

For a comparison between the new and previous versions of Intelligent Media Management, see New and old version guide.
For more information about image processing in the new version of Intelligent Media Management, see Image processing.

Notes

The JPG, PNG, and BMP image formats are supported. GIF images are also supported for all features except for QR code detection. For GIF images, detection is performed on the first static frame.
For WEBP, TIFF, and HEIC images, Intelligent Media Management automatically attempts to convert the image format using the OSS image transform feature before performing AI-based image processing.
Important
This process may incur conversion fees. For more information, see Data processing billing.
The maximum size for an input image is 100 MB. For the QR code detection feature, the minimum image size is 40 px × 40 px.
There is no limit on the number of faces that can be detected in an image. By default, all detected faces are returned. The detection time depends on the image size, with larger images taking longer to process.
Requests for very large images may time out. If a timeout occurs, reduce the image size and try again.

Features

Name	Description
Content detection	Detects labels in images. For more information, see Content detection.
QR code detection	Detects QR codes in images and returns their content and location. For more information, see QR code detection.
Image crop suggestion	Provides crop suggestions for an image based on a specified aspect ratio. For more information, see Image crop suggestion.
Image quality scoring	Detects the overall quality score of an image. For more information, see Image quality scoring.
Face detection	Detects faces in images. For more information, see Face detection.
Face search	Searches a media set for the top N images that are most similar to a specified image or face ID. For more information, see Face search.
Face comparison	Compares the similarity of two faces. This can be done in two ways. For more information, see Face comparison. Without creating a media set, directly pass the OSS URIs of two images to compare the similarity of the largest face in each image. Create a media set, index the images into the set using the IndexImage operation, and then pass two FaceIds from the media set to compare them.
Face clustering	Groups images in a media set that contain similar faces. After clustering, use the face group to query for all images of a specific person. For more information, see Face clustering.
Body detection	Detects the location of human bodies in an image. For more information, see Body detection.
Blind image watermark	Adds a blind watermark to an image. For more information, see Blind image watermark. After a blind watermark is added, it is not directly visible. To recover the watermark information, use the blind watermark parsing feature.
Tag search	Create a media set and index images into it. Then, search for images in the media set by time, location, tag, or person feature. For more information, see Tag search.