Predictor

A room type predictor using tensorflow. Currently we predict based on the content type of each content exists in the room.

Setup & Run - Live Demo

To run the live demo, you need to execute the following command.

$ cd server
$ pip install -r requirements.txt
$ python app.py

Setup & Run - Classifier

$ cd tensorf
$ python roomtype-softmax.py

Structure Introduction

/designjsons

This folder contains the input data for our predicting algorithm. Each file is a design json with the following format:

{
  "1435" {                                # the room id
    "category": "LivingDiningRoom",       # room type of this room
    "id": "1435",
    "characters": [                       # list of contents that belongs to this room
      {                          
        "id": "e8ad5df2-2e4c-4d6e-9174-2b6bb55fdc38",     # content id/seekid
        "contenttype": "sofa/double seat sofa"            # content type
      },
      {
        "id": "48512a7c-b04d-4487-ae45-bf513e684556",
        "contenttype": "sofa/double seat sofa"
      },
      ...
 },
  ...
}

/dwgjsons

This folder contains the original design jsons which are exported from floorplan.

/server

This folder contains the code for live demo.

/tensorf

Predictor algorithm.

/validators

Data used for validation. Same format as the ones in /designjsons.

iratao/predictor