Updated main file

.gitignore

@@ -5,4 +5,7 @@ output_patch_temp/
 output_test_*/
 weights/
 fiji/
-*.csv
+*.csv
+data/
+__pycache__
+output_patch_jpg/

README.md

@@ -39,7 +39,7 @@ python main.py --pilot=0
 Installation option 2:
 first to install a PyImageJ environment with all packages satisfied
 conda install mamba -n base -c conda-forge
-mamba create -n pyimagej -c conda-forge pyimagej openjdk=8
+mamba create -n pyimagej -c conda-forge pyimagej openjdk=11
 conda activate pyimagej
 Then, inside this environment, install dependencies for [PyTorch](https://pytorch.org/) and a bunch of other image processing packages such as sciki-image.
 

convert.py

@@ -0,0 +1,26 @@
+import os
+from PIL import Image
+
+yourpath = os.getcwd() + '/output_patch_temp/'
+outpath = os.getcwd() + '/output_patch_jpg/'
+
+def convert_to_jpeg(img_name):
+    yourpath = os.getcwd() + '/output_patch_temp/'
+    outpath = os.getcwd() + '/output_patch_jpg/' + img_name
+
+    for root, dirs, files in os.walk(yourpath, topdown=False):
+        for name in files:
+            #print(os.path.join(outpath, name))
+            if os.path.splitext(os.path.join(root, name))[1].lower() == ".tiff":
+                if os.path.isfile(os.path.splitext(os.path.join(outpath, name))[0] + ".jpg"):
+                    print(f"A jpeg file already exists for {name}")
+                # If a jpeg is *NOT* present, create one from the tiff.
+                else:
+                    outfile = os.path.splitext(os.path.join(outpath, name))[0] + ".jpg"
+                    try:
+                        im = Image.open(os.path.join(root, name))
+                        print("Generating jpeg for %s" % name)
+                        im.thumbnail(im.size)
+                        im.save(outfile, "JPEG", quality=100)
+                    except(Exception, e):
+                        print(e)

env.yml

@@ -5,12 +5,11 @@ channels:
   - conda-forge
   - defaults
 dependencies:
-  - python=3.7.6
-  - cudatoolkit=10.2
+  - cudatoolkit
   - numpy
   - pillow
   - pip
-  - pytorch=1.8.0
+  - pytorch
   - torchvision
   - matplotlib
   - pyyaml
@@ -19,7 +18,7 @@ dependencies:
   - tqdm
   - imglyb
   - jpype1
-  - openjdk=8
+  - openjdk=11
   - pyimagej
   - scyjava
   - xarray

main.py

@@ -1,11 +1,10 @@
 from skimage import io, img_as_ubyte, morphology, img_as_bool, img_as_float, exposure, color
 from skimage.util.shape import view_as_windows
-from skimage.util import crop, pad
+from skimage.util import crop
 from skimage.transform import resize, rescale
 from PIL import Image
 import imagej
 
-
 import numpy as np
 import os, glob, sys
 from collections import OrderedDict
@@ -24,6 +23,28 @@
 
 import model_SHG as md
 
+def convert_to_jpeg(img_name):
+    yourpath = os.getcwd() + '/output_patch_temp/'
+    outpath = os.getcwd() + '/output_patch_jpg/' + img_name
+    os.makedirs(f'output_patch_jpg/{img_name}/')
+    print("Generating jpeg for %s" % img_name)
+
+    for root, dirs, files in os.walk(yourpath, topdown=False):
+        for name in files:
+            #print(os.path.join(outpath, name))
+            if os.path.splitext(os.path.join(root, name))[1].lower() == ".tiff":
+                if os.path.isfile(os.path.splitext(os.path.join(outpath, name))[0] + ".jpg"):
+                    print(f"A jpeg file already exists for {name}")
+                # If a jpeg is *NOT* present, create one from the tiff.
+                else:
+                    outfile = os.path.splitext(os.path.join(outpath, name))[0] + ".jpg"
+                    try:
+                        im = Image.open(os.path.join(root, name))
+                        im.thumbnail(im.size)
+                        im.save(outfile, "JPEG", quality=100)
+                    except(Exception, e):
+                        print(e)
+
 def generate_csv(img_dir, csv_dir):
     file_list= [name for name in os.listdir(img_dir) if 
                 os.path.isfile(os.path.join(img_dir, name))]
@@ -104,7 +125,8 @@ def demo(args):
     model.to(device)
 
     print('loading ImageJ, please wait')
-    ij = imagej.init('fiji/Fiji.app/')
+    # ij = imagej.init('fiji/Fiji.app')
+    ij = imagej.init('sc.fiji:fiji:2.1.1')
 
     # use for SHG
     TASK = args.input_folder
@@ -138,7 +160,7 @@ def demo(args):
         canvas_1 = int(window_shape[1] * shape_1_factor)
         pad_0 = canvas_0 - img.shape[0]
         pad_1 = canvas_1 - img.shape[1]
-        canvas = pad(img, ((0, pad_0), (0, pad_1), (0, 0)), mode='reflect')
+        canvas = np.pad(img, ((0, pad_0), (0, pad_1), (0, 0)), mode='reflect')
         windows = view_as_windows(canvas, window_shape, step_size)
         with open(OUTPUT_PATCH_DIR+'TileConfiguration.txt', 'w') as text_file:
             print('dim = {}'.format(2), file=text_file)
@@ -184,6 +206,7 @@ def demo(args):
         c1 = exposure.rescale_intensity(c1, in_range=(0, 255), out_range=(0, 1))
         print(str(k+1)+"/" + str(len(files)) + " output saved as: " + output_name)
         io.imsave(output_name, img_as_ubyte(c1))
+        convert_to_jpeg(fn)
         if args.pilot:
             break
 

model_SHG.py

@@ -3,6 +3,12 @@
 import torch.nn.init as init
 import torch.nn.functional as F
 import math
+from torch.utils.data import Dataset, DataLoader
+from torchvision import utils
+import torch.functional as F
+import torch
+
+import model_SHG as md
 
 class SkipBlock(nn.Module):
     def __init__(self, in_features, out_features):

packages.txt

@@ -1,7 +1,10 @@
-matplotlib==3.1.2
-numpy==1.17.4
-pandas==0.25.3
-Pillow==5.3.0
-scikit-image==0.16.2
-tqdm==4.42.0
-pytorch==1.3.1
+matplotlib
+numpy
+pandas
+Pillow 
+scikit-image
+tqdm
+torch
+imagej
+torchvision
+pyimagej

train.py

@@ -0,0 +1,127 @@
+import torch
+import torch.nn as nn
+import torch.nn.init as init
+import torch.nn.functional as F
+import math
+from skimage import io, img_as_ubyte, morphology, img_as_bool, img_as_float, exposure, color
+from skimage.util.shape import view_as_windows
+from skimage.util import crop
+from skimage.transform import resize, rescale
+from PIL import Image
+import imagej
+
+import numpy as np
+import os, glob, sys
+from collections import OrderedDict
+import shutil
+import argparse
+import pandas as pd
+import csv
+import warnings
+warnings.simplefilter("ignore", UserWarning)
+
+
+from torch.utils.data import Dataset, DataLoader
+from torchvision import utils
+import torch.functional as F
+import torch
+
+import model_SHG as md
+
+"""Fazer Augmentation e arrumar funções de Loss"""
+model = md.GeneratorUNet(3,1)
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model = model.to(device)
+
+
+num_epochs = 250
+best_loss = float('inf')
+best_loss_ce = float('inf')
+best_loss_dice = float('inf')
+best_iou_score = float('inf')
+best_accuracy = float('inf')
+
+patience = 25
+trigger_times = 0
+accumulation_steps = 4
+
+
+for epoch in range(num_epochs):
+    model.train()
+    train_loss, train_ce_loss, train_dice_loss = 0.0, 0.0, 0.0
+    train_iou_score, train_accuracy = 0.0, 0.0
+    # contador de iteracao
+    i_loop = 0
+    for images, masks in train_loader:
+        images, masks = images.to(device), masks.to(device)
+
+        outputs = model(images)
+        # outputs = F.interpolate(outputs, size=(256, 256), mode='bilinear', align_corners=False)
+        loss, ce_loss, dice_loss = criterion(outputs, masks)
+        loss.backward()
+
+        iou_score = m_iou(outputs, masks)
+        accuracy = pixel_accuracy(outputs, masks)
+
+        if (i_loop + 1) % accumulation_steps == 0:
+            optimizer.step()
+            optimizer.zero_grad()
+
+        i_loop +=1
+        train_loss += loss.item() * images.size(0)
+        train_ce_loss += ce_loss.item() * images.size(0)
+        train_dice_loss += dice_loss.item() * images.size(0)
+        train_iou_score += iou_score * images.size(0)
+        train_accuracy += accuracy * images.size(0)
+
+    train_loss /= len(train_loader.dataset)
+    train_ce_loss /= len(train_loader.dataset)
+    train_dice_loss /= len(train_loader.dataset)
+    train_iou_score /= len(train_loader.dataset)
+    train_accuracy /= len(train_loader.dataset)
+
+    model.eval()
+    valid_loss, valid_ce_loss, valid_dice_loss = 0.0, 0.0, 0.0
+    valid_iou_score, valid_accuracy = 0.0, 0.0
+    with torch.no_grad():
+        for images, masks in valid_loader:
+            images, masks = images.to(device), masks.to(device)
+            outputs = model(images)
+            # outputs = F.interpolate(outputs, size=(256, 256), mode='bilinear', align_corners=False)
+
+            loss, ce_loss, dice_loss = criterion(outputs, masks)
+            iou_score = m_iou(outputs, masks)
+            accuracy = pixel_accuracy(outputs, masks)
+
+
+            valid_loss += loss.item() * images.size(0)
+            valid_ce_loss += ce_loss.item() * images.size(0)
+            valid_dice_loss += dice_loss.item() * images.size(0)
+            valid_iou_score += iou_score * images.size(0)
+            valid_accuracy += accuracy * images.size(0)
+
+    valid_loss /= len(valid_loader.dataset)
+    valid_ce_loss /= len(valid_loader.dataset)
+    valid_dice_loss /= len(valid_loader.dataset)
+    valid_iou_score /= len(valid_loader.dataset)
+    valid_accuracy /= len(valid_loader.dataset)
+
+    print(f"Epoch {epoch+1}/{num_epochs}, Train Loss: {train_loss:.3f} (CE: {train_ce_loss:.3f}, Dice: {train_dice_loss:.3f}, IoU: {train_iou_score:.3f}, Pixel Accuracy: {train_accuracy:.3f}). Valid Loss: {valid_loss:.3f} (CE: {valid_ce_loss:.3f}, Dice: {valid_dice_loss:.3f}, IoU: {valid_iou_score:.3f}, Pixel Accuracy: {valid_accuracy:.3f})")
+
+    # Early stopping
+    if valid_loss < best_loss:
+        best_loss = valid_loss
+        best_loss_ce = valid_ce_loss
+        best_loss_dice = valid_dice_loss
+        best_iou_score = valid_iou_score
+        best_accuracy = valid_accuracy
+        torch.save(model.state_dict(), 'best_model.pth')
+        trigger_times = 0
+    else:
+        trigger_times += 1
+        if trigger_times >= patience:
+            print('Early stopping!')
+            break
+
+print("Training complete.")