Merge branch 'develop-freek' into 'develop'

Update code/4. Data loader.ipynb See merge request !6

Merge branch 'develop-freek' into 'develop'
dc2808c8 · Freek Daniëls · 947f57c0 · 438e6978 · dc2808c8
Commit dc2808c8 authored 2 years ago by Freek Daniëls
--- a/code/4. Data loader.ipynb
+++ b/code/4. Data loader.ipynb
@@ -75,7 +75,7 @@
    "                    annotation = {}\n",
    "                    annotation['name'] = name\n",
    "                    annotation['configuration'] = configuration\n",
-    "                self.annotations.append(annotation)\n",
+    "                    self.annotations.append(annotation)\n",
    "      \n",
    "        self.transforms = transforms        \n",
    "\n",

 %% Cell type:code id: tags:

 ``` python
 from __future__ import print_function
 from __future__ import division

 import pandas as pd
 import numpy as np

 import glob
 import os
 import random
 import time
 import copy
 import json

 from dataclasses import dataclass

 from PIL import Image

 from skimage import io

 import matplotlib.pyplot as plt

 import torch.utils.data
 import torch
 import torchvision.transforms.functional as TF

 from IPython.display import clear_output
 ```

 %% Cell type:code id: tags:

 ``` python
 %run "0. Config.ipynb"
 %run "1. Utils.ipynb"
 ```

 %% Cell type:code id: tags:

 ``` python
 class SeagrassDataset(torch.utils.data.Dataset):

    def __init__(self, path_to_images, path_to_masks, path_to_configurations, path_to_files, transforms=False):

        self.path_to_images = path_to_images
        self.path_to_masks = path_to_masks
        self.annotations = []

        files = pd.read_csv(path_to_files,usecols=["file"])

        for row in files['file']:

            configurations = path_to_configurations + row + ".json"

            with open(configurations) as json_file:
                data = json.load(json_file)

                name = data['name']

                for configuration in data['configurations']:
                    annotation = {}
                    annotation['name'] = name
                    annotation['configuration'] = configuration
-                self.annotations.append(annotation)
+                    self.annotations.append(annotation)

        self.transforms = transforms

    def __getitem__(self, idx):

        annotation = self.annotations[idx]

        name = annotation['name']
        configuration = annotation["configuration"]
        configuration_mask = configuration["mask"]


        mask_left = configuration_mask['left']
        mask_right = configuration_mask['right']
        mask_top = configuration_mask['top']
        mask_bottom = configuration_mask['bottom']

        image = io.imread(self.path_to_images + name)
        mask = io.imread(self.path_to_masks + name)

        if self.transforms:
            image = pad_image(image,image_padding_for_augmentation)
            mask = pad_image(mask,mask_padding_for_augmentation)
            image_offset = image_padding_for_augmentation
            mask_offset = mask_padding_for_augmentation
        else:
            image = pad_image(image,image_padding)
            image_offset = image_padding
            mask_offset = 0

        image_left = mask_left-image_offset+image_offset
        image_top = mask_top-image_offset+image_offset
        image_right = mask_right+image_offset+image_offset
        image_bottom = mask_bottom+image_offset+image_offset

        mask_left = mask_left-mask_offset+mask_offset
        mask_top = mask_top-mask_offset+mask_offset
        mask_right = mask_right+mask_offset+mask_offset
        mask_bottom = mask_bottom+mask_offset+mask_offset

        image = Image.fromarray(image).convert("RGB")
        mask = Image.fromarray(mask)

        if self.transforms:

            image = image.crop((image_left,image_top,image_right,image_bottom))
            mask = mask.crop((mask_left,mask_top,mask_right,mask_bottom))

            if random.random() > 0.5:
                image = TF.hflip(image)
                mask = TF.hflip(mask)

            if random.random() > 0.5:
                image = TF.vflip(image)
                mask = TF.vflip(mask)

            degrees = random.random()*180
            image = TF.affine(image,degrees,translate=[0,0], scale=1, shear=0.0)
            mask = TF.affine(mask,degrees,translate=[0,0], scale=1, shear=0.0)

        image = TF.center_crop(image,image_size)
        mask = TF.center_crop(mask,mask_size)


        image = TF.to_tensor(image)

        image = TF.normalize(image,mean,std)

        mask = torch.tensor(np.array(mask)).long()

        return image, mask, name

    def __len__(self):
        return len(self.annotations)
 ```

 %% Cell type:code id: tags:

 ``` python
 datasets = {"train":SeagrassDataset(path_to_images,path_to_masks,path_to_configurations,training_file,True),
            "val":SeagrassDataset(path_to_images,path_to_masks,path_to_configurations,validation_file),
            "test":SeagrassDataset(path_to_images,path_to_masks,path_to_configurations,test_file)}
 ```

 %% Cell type:code id: tags:

 ``` python
 image,mask,name = next(iter(datasets["train"]))
 image = np.transpose(np.asarray(image),axes=[1,2,0])*std+mean

 mask = np.asarray(mask)

 plt.figure(figsize=(10,5))
 plt.subplot(1, 2, 1)
 plt.axis('off')
 plt.tight_layout()
 plt.imshow(image)

 plt.subplot(1, 2, 2)
 plt.axis('off')
 plt.tight_layout()
 plt.imshow(get_coloured_mask(mask,labels))

 plt.show()
 ```