Skip to content

CS 5542 BigData Lab Report #11

Jump to bottom
Amy Lin edited this page Apr 12, 2017 · 7 revisions

TensorFlow Show & Tell Captions of images

[ QUESTION ]

4 captions for each images of 3 Kansas City Fountains ( Children's Fountain + J.C. Nichols Memorial Fountain + Muse of Missouri ).

[ IMPLEMENTATION ]

  • Get all the images from the data sets.
  • Build the inference graph.
  • Create vocabularies from the vocab_file.
  • Load the model for checkpoints.
  • Write 4 captions of each images to fountainCaptions.txt file.

< DATASET >

3 Kansas City Fountains

[ RESULTS ]

Captions for each image

Web-based Q & A

[ QUESTION ]

Web-based system for answering questions.

[ IMPLEMENTATION ]

  • Deploy Tensorflow onto Heroku.
  • Use Heroku CLI deploy plugin to connect Heroku-CLI to my own account.
  • After deployment, we can check if the app name is on Heroku.
  • Import project onto Api.ai
  • Build UI locally through React-app and starting npm.
  • Open browser and google conversation to start using application!

[ RESULTS ]

I was stuck after I created the Heroku-CLI and am still trying to connect everything up!

Clone this wiki locally