Microsoft Cognitive Services Development Practice (1)-Oxford Program Introduction _ Machine Learning

Brief introduction

In recent years, because of the cloud platform, large data, high-performance computing, machine learning and other areas of progress, artificial intelligence also fire up. Face recognition, speech recognition and other related functions have been proposed, but can form products and large-scale use of small. Because it is difficult for non-professional professionals to achieve a complete set of artificial intelligence program, involving artificial intelligence can only find open source of the library, after all, so they go to training the network, their own to learn the use of various libraries and transplant to their own programs.

In the implementation of some products, the use of artificial intelligence than the development of artificial intelligence is more important, so Microsoft proposed the "Oxford Plan" (Oxford Project), the purpose is to use artificial intelligence to do some function developers can focus on how to use such as face recognition, Speech recognition and other functions of the business logic, rather than how to implement the face recognition, speech recognition and other algorithms in detail. It is serviced through its Azure cloud platform and is currently free of charge for a small number of requests.

The goal of the Oxford program is to get developers to focus on the product itself rather than the technical details. And Microsoft has provided a professional platform service through the Oxford program. Oxford Program (Oxford Project) features

Microsoft Oxford plans to provide three of artificial intelligence solutions in the field of direction, respectively, "vision", "voice" and "language."
The following links are the homepage of the Oxford Program (Oxford Project):

https://www.projectoxford.ai/

In the Oxford program, Microsoft offers a common use of the mainstream AI services in 10 directions, "vision", "voice" and "language". Currently free use (or trial) for less traffic. This article will briefly introduce the free limit and the charging standard of the charging service. Visual

Oxford plans to provide 4 services in the field of visual intelligence. Computer Vision (Computer Vision API)

Microsoft Oxford plans to provide the following services in the computer Vision direction: Image detection Analysis OCR optical character recognition generate thumbnails

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/vision facial recognition (face API)

Microsoft Oxford plans to provide the following services in the direction of face recognition: face detection with face verification similar face search face Group face recognition

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/face Emotion Recognition (video API)

Microsoft Oxford plans to provide services by uploading a photograph by giving the results of "anger", "contempt", "disgust", "fear", "happiness", "neutral", "sadness", and "surprise" by scoring them separately.

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/Emotion video detection (face API)

Microsoft Oxford plans to provide the following services in the direction of video detection: Video stability processing face detection and tracking moving object detection

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/video Voice

Oxford plans to provide 3 services in the field of voice intelligence. Speech recognition (Speech API)

Microsoft Oxford plans to provide the following services in the direction of speech recognition: Voice text (speech recognition, support for Chinese, you can do a Siri with semantic detection and speech synthesis) semantic detection speech synthesis (TTS)

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/speech voice Recognition (Speaker recognition API)

Microsoft Oxford plans to provide the following services in the voice print recognition direction: Voice print verification (generally used to verify user identity) speech recognition (generally used to automatically identify the members of the current sound involved)

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/speech CRIS (custom speech recognition service)

Microsoft's Oxford program provides a custom service for speech recognition, which constructs a custom model and provides services through a custom language model, voice model. I have never studied this, so I do not make introductions and comments. Language

Oxford plans to provide 3 services in the field of language intelligence. Spell check and correction (Spell check API)

The spell checker service is similar to the spelling checker in Word, because it is a cloud detection that makes more accurate checks based on current buzzwords and new words. Compared with the test in Word, I think the inspection and correction are more accurate.

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/SpellCheck LUIS (Language understanding Service)

Language understanding services can be understood as Apple's Siri or Microsoft's Cortana. To understand the semantics of specific statements by automatically interpreting the verb names in the input language. WEBLM API

I don't know why this is called the Web language model, and perhaps the content involved is more relevant to the Web. This part of the Oxford plan is important to identify the probability of a word appearing, and the probability of the word appearing at a given position, and ultimately to use the probability information to do the participle. (N-gram algorithm, the author personally is so understanding.) ）

The relevant demo address is as follows:

Https://www.projectoxford.ai/demo/WebLM Summary

This article, as the first of the Microsoft "Oxford Program" series, focuses on the services currently offered, and hopes that the reader can understand the current "Oxford program" to provide the services you need, based on the above summarized presentation. And according to the specific services, to retrieve documents and apply.

The author is also using some of the "Oxford Program" services to develop some small applications for future articles to show how to use the specific API.