AI Engines Configurations

This section outlines the setup requirements for various AI engines, including Tesseract.js, TensorFlow.js, and NLP.js. Each configuration specifies the necessary parameters to enable the models managed by the respective AI engines.

Tesseract.js

Tesseract.js is a JavaScript library that brings the capabilities of the Tesseract OCR engine to the web by compiling it from C to JavaScript using WebAssembly. This enables optical character recognition (OCR) to be performed directly in the browser, facilitating faster text recognition and minimizing the need for server-side processing.

Tesseract.js supports over 100 languages, automatic text orientation, and script detection, making it a versatile tool for extracting text from images and videos in various languages and formats.

Primary Parameters:

Parameter	Type	Default Value	Required	Description	Metal Version
engine	String		Yes	Set to tesseractjs for Tesseract.js	^0.1
model	String	eng	No	Language to use for OCR (see Lang Code: Tesseract OCR Data Files)	^0.1

Example Configuration:

ai-engines:
  my-ocr:
    engine: tesseractjs
    model: eng

Output Object:

Key	Type	Description
blocks	Array<Block>	An array of text blocks detected in the image.
confidence	Number	The overall confidence level of the recognized text.
html	String	The recognized text formatted as HTML.
lines	Array<Line>	An array of lines detected in the image.
oem	String	OCR engine mode used during OCR processing.
paragraphs	Array<Paragraph>	An array of paragraphs detected in the image.
psm	String	Page segmentation mode used during OCR processing.
symbols	Array<Symbol>	An array of individual symbols detected in the image.
text	String	The recognized text from the image.
version	String	The version of the OCR engine.
words	Array<Word>	An array of individual words detected in the image.

TensorFlow.js

TensorFlow.js is an open-source JavaScript library that allows you to develop machine learning and deep learning models directly in the browser. It is part of the TensorFlow ecosystem and enables running machine learning models, including neural networks, in JavaScript.

Primary Parameters:

Parameter	Type	Required	Description	Metal Version
engine	String	Yes	Set to tensorflowjs for TensorFlow.js	^0.1
model	String	Yes	Model to use (see: Models table)	^0.1

Models:

Model Value	Description	Metal Version
image-classify	To classify image	^0.1

image-classify

Image classification involves identifying and categorizing images into predefined classes.

Example Configuration:

ai-engines:
  my-image-classifier:
    engine: tensorflowjs
    model: image-classify

Output Object:

{
  "class": [
    {
      "className": "screw",
      "probability": 0.6480134725570679
    },
    {
      "className": "acoustic guitar",
      "probability": 0.31899121403694153
    },
    {
      "className": "electric guitar",
      "probability": 0.014988371171057224
    }
  ]
}

Key	Type	Description
class	Array	An array of classification results
class[].className	String	The name of the classified object
class[].probability	Float	The probability of the classified object

NLP.js

NLP.js is an open-source JavaScript library for natural language processing (NLP) and natural language understanding (NLU). It provides tools for various NLP tasks such as tokenization, part-of-speech tagging, named entity recognition, and sentiment analysis.

Primary Parameters:

Parameter	Type	Required	Description	Metal Version
engine	String	Yes	Set to nlpjs for NLP.js	^0.1
model	String	Yes	Model to use (see: Models table)	^0.1

Models:

Model Value	Description	Metal Version
sentiment	For sentiment analysis	^0.1
lang-guess	For guessing the language	^0.1

sentiment

Sentiment analysis determines the sentiment expressed in a given text, typically categorizing sentiments as positive, negative, or neutral. This process is essential for understanding customer feedback, monitoring social media, and analyzing opinions and emotions in various applications.

Optional Parameters:

Parameter	Type	Default Value	Required	Description	Metal Version
lang	String	en	No	Defines the language of the text to be analyzed. For supported languages, see NLP.js Sentiment Analysis Languages.	^0.1

Example Configuration:

ai-engines:
  my-sentiment-analyzer:
    engine: nlpjs
    model: sentiment
    options:
      lang: en

Output Object:

{
  "score": 1.125,
  "numWords": 4,
  "numHits": 3,
  "comparative": 0.28125,
  "type": "senticon",
  "language": "en"
}

Key	Type	Description
score	Float	The overall sentiment score
numWords	Integer	The total number of words analyzed
numHits	Integer	The number of words that were found in the sentiment lexicon
comparative	Float	The comparative score, calculated as score/numWords
type	String	The type of sentiment analysis method used
language	String	The language of the analyzed text

lang-guess

Language guessing can identify the language of a given text. This feature is useful for applications requiring language detection before processing text, such as multilingual chatbots, content categorization, and automatic language selection. It ensures that the appropriate language models and processing techniques are applied, enhancing the accuracy and relevance of the application's responses.

Optional Parameters:

Parameter	Type	Default Value	Required	Description	Metal Version
accept	String	en	No	Defines a list of languages to guess, separated by commas. For supported languages, see NLP.js Language Support.	^0.1

Example Configuration:

ai-engines:
  my-language-guesser:
    engine: nlpjs
    model: guess-lang
    options:
      accept: en,fr

Output Object:

The returned object is an array of guessed languages with their respective scores:

[
  {
    "alpha3": "eng",
    "alpha2": "en",
    "language": "English",
    "score": 1
  }
]

Key	Type	Description
alpha3	String	The three-letter ISO 639-2 code for the language
alpha2	String	The two-letter ISO 639-1 code for the language
language	String	The name of the language
score	Float	The confidence score for the language guess