This topic describes the 2022 product and documentation updates for the Alibaba Cloud Visual Intelligence API.
December 2022
Category | Capability Name | Description | Release date | Supported platforms | References |
Offline SDK | Server-side facial recognition offline SDK | This SDK supports face detection, face tracking, face alignment, face quality assessment, pose estimation, liveness detection, and facial recognition. You can deploy it directly on a server. The SDK includes an authorization feature. After successful authorization, it works offline. You can customize development based on your business needs. | 2022-12-30 | Linux server |
Category | Capability Name | Description | Release date | Region | References |
Video production | Template-based video face swapping | After obtaining user consent, this capability detects the largest face in a video and replaces it with another person's facial features to achieve a face-swapping effect. | 2022-12-02 | China (Shanghai) | |
Add video face swap templates | This capability lets you add videos that contain faces and have passed content moderation as templates for template-based video face swapping. | 2022-12-02 | China (Shanghai) | ||
Query video face swap templates | This capability lets you query video face swap templates that you have added. | 2022-12-02 | China (Shanghai) | ||
Delete video face swap templates | This capability lets you delete video face swap templates that you have added. | 2022-12-02 | China (Shanghai) | ||
Face and human body | Face liveness detection | This upgraded face liveness detection algorithm improves the blocking rate for spoofing attacks such as photo replay, image editing, printed photos, and high-fidelity molds. It works across devices including smartphones, access control cards, attendance machines, and PCs. | 2022-12-09 | China (Shanghai) | |
Face comparison (1:1) | This upgraded 1:1 facial recognition algorithm compares the largest face from each of two authorized images to determine whether they belong to the same person. It returns bounding box coordinates for both faces, the confidence level of the comparison, and confidence thresholds for different false acceptance rates. The algorithm achieves over 99% accuracy. | 2022-12-30 | China (Shanghai) |
September 2022
Category | Capability Name | Description | Release date | Region | References |
Image analysis and processing | Esophageal cancer detection | This capability assesses esophageal cancer risk using input chest CT scans. It works with any non-contrast CT scan that covers the esophagus, such as chest or abdominal CT scans. | 2022-09-20 | China (Shanghai) |
August 2022
Category | Capability Name | Description | Release date | Region | References |
Image analysis and processing | Session feedback | In dermatology pre-consultation scenarios, a session includes multiple rounds of Q&A. This interface collects feedback after the session ends to support additional data transmission, such as custom data defined by the caller. | 2022-08-31 | China (Shanghai) |
July 2022
Category | Capability Name | Description | Release date | Region | References |
Image analysis and processing | Multi-organ segmentation | This capability identifies and segments organs at risk in radiotherapy scenarios using input chest CT images. | 2022-07-19 | China (Shanghai) |
June 2022
Category | Capability Name | Description | Release date | Region | References |
Video understanding | Video OCR | This capability recognizes text in videos. It supports Chinese and English text, simplified and traditional characters, scores, and more. It works across news, TV shows, entertainment, sports, and other scenarios. It handles standard captions, static captions, scrolling captions, natural scene text, vertical text, and stylized fonts. | 2022-06-29 | China (Shanghai) |
May 2022
Category | Capability Name | Description | Release date | Region | References |
Image analysis and processing | Pancreatic cancer detection | This capability assesses pancreatic cancer risk using input chest CT scans. | 2022-05-19 | China (Shanghai) | |
Lymph node detection | This capability detects enlarged lymph nodes in chest CT scans, including mediastinal, hilar, and supraclavicular lymph nodes. | 2022-05-19 | China (Shanghai) |
April 2022
Category | Capability Name | Description | Release date | Region | References |
Video understanding | Video splitting | This capability splits videos into segments along multiple dimensions, such as shot, person, subject, and scene. It also generates summaries for each segment. | 2022-04-30 | China (Shanghai) |
March 2022
Category | Capability Name | Description | Release date | Region | References |
OCR | Video text recognition | This capability performs structured processing on input videos and returns text content, text region coordinates, and timestamps. | 2022-03-30 | China (Shanghai) |