plugin: help us to implement model plugins

[yorkie]

We have the following categories that we need to support at Pipeline and MLApp.

Working with Image Data

• Image Classification (it does classify the image to specific classes)
  • model available: ResNet by @utkobe add resnet tensorflow model #186
• Image Generation (it generates image automatically)
  • model available: CycleGAN
• Object Detection (it detects the given objects and returns class and position for each one)
  • model available: Faster-RCNN
• Image Segmentation (it returns the outline of given objects in pixel)
  • model available: Master-RCNN @utkobe
• Image Clustering (it returns clusters from images)
  • model available: ClusterCNN @WenheLI

Working with Text Data

• Text Classification (it does classify the text to specific classes)
  • model available
    • FastText @yorkie model: add new fasttext plugins and pipeline #216
    • TextRNN
• Named Entity Recognition (it recognizes the named entity from a sentence)
  • model available: BiLSTM + CRF
• Relationship Extraction (it extracts the relationships between sentences)
  • model available: Bert
• Coreference Resolution (it does the pronouns and other referrings to be connected with right individuals.)
• Writing Correction (it helps to do writing correction)
  • model available: Bert
• Machine Translation (it translates a language to target language)
  • model available: Transformer
• Question and Answering (it generates an answer by given question)
  • model available: Bert
• Text Summary (it generates a summary for your given long text)
  • model available: pointer-generator-network @gindis
• Text Creation (it generates an artwork by given portfolio)
  • model available: LSTM @yorkie
• Text Clustering (it returns clusters from words or sentences)
  • model available: CNN, RNN

If you are interested in how to develop a model plugin and write a new Pipeline, feel free to try with the above and open a new pull request :)

[Txiaozhe]

awesome~