GCNet-CCVNorm

GCNet_CCVNorm (revised)

This code is revised from the official code, only the input and output is changed.

Introduction

Because the GCNet doesn't provide the official code, the base method is GCNet, so in the paper, we use this version of GCNet.

Example

GCNet

Testing

We will give an example in the code folder, the file is shown in example folder, the pre-trained model is in pre-trained folder.

We can test the example by running :

#! /bin/bash

./evaluate_color_example_GCNet.sh

The result can be shown :

The result of GCNet on DublinCity

GCNet_CCVNorm

Testing

We will give an example in the code folder, the file is shown in example folder, the pre-trained model is in pre-trained folder.

We can test the example by running :

#! /bin/bash

./evaluate_color_example_GCNet_CCVNorm.sh

In the example, the guidance is 0.5%, the result can be show :

The result of GCNet-CCVNorm on DublinCity

Training

The training step depends on the structure of the training data, we will git an example in the file train_toulouse_guide.sh.

Feed Back

If you think you have any problem, contact Teng Wu [email protected]