Skip to content
A vertex prediction model implementation based on
Branch: master
Clone or download
Latest commit e346ef1 Sep 8, 2018
Type Name Latest commit message Commit time
Failed to load latest commit information.

TensorFlow implementation of vertex prediction model:

  1. The following repository includes a vertex prediction model that may be used to infer mappings from objects in images to the vertices of the polygon which bound the object of interest.
  2. Using a modified VGG-16 convolutional neural network that takes as input a 224x224x3 RGB image, inference is done in one forward pass that happens in two stages. First, an image representation is extracted from blocks of the VGG-16 model. Then a down-sampled variant of this image-representation is passed to a fully-connected layer which performs a global optimisation step.
  3. For more details concerning the reference paper, I would recommend reading 'Annotating Object Instances with a Polygon-RNN': as well as:
You can’t perform that action at this time.