There are a few helper functions I want to highlight in this section in the following steps:

1. The load data function helps define the downloaded input data—the images are stitched together as 256 x 512, so this function reads them in and splits them into two arrays: 

def load_data(self,data_path):
    listOFFiles = self.grabListOfFiles(data_path,extension="jpg")
    imgs_temp = np.array(self.grabArrayOfImages(listOFFiles))
    imgs_A = []
    imgs_B = []
    for img in imgs_temp:
        imgs_A.append(img[:,:self.H])
        imgs_B.append(img[:,self.H:])

     imgs_A_out = self.norm_and_expand(np.array(imgs_A))
     imgs_B_out = self.norm_and_expand(np.array(imgs_B))

     return imgs_A_out, imgs_B_out

2. This is a convenience function that puts the array in the correct shape for the network to use it:

def norm_and_expand(self,arr):
    arr = (arr.astype(np.float32) - 127.5)/127.5
    normed = np.expand_dims(arr, axis=3)
    return normed

3. This function lets us grab a list of files from a starting directory:

def grabListOfFiles(self,startingDirectory,extension=".webp"):
    listOfFiles = []
    for file in os.listdir(startingDirectory):
        if file.endswith(extension):
           listOfFiles.append(os.path.join(startingDirectory, file))
        return listOfFiles

4. Given a list of files, read in the images into an array and return them:

def grabArrayOfImages(self,listOfFiles,gray=False):
    imageArr = []
    for f in listOfFiles:
        if gray:
           im = Image.open(f).convert("L")
        else:
           im = Image.open(f).convert("RGB")
       imData = np.asarray(im)
       imageArr.append(imData)
    return imageArr

Now, we'll move on to how we train this model now that we have the class setup and working.