standardize image size, add requirement, add readme, add image sample

.gitignore

@@ -0,0 +1 @@
+venv

Code/classification.py

@@ -2,7 +2,7 @@
 import numpy as np
 import pickle
 class inference_glaucoma():
-  def __init__(self, model_path='Code/Models/Inference model'):
+  def __init__(self, model_path='Models/Inference model/inference_model.sav'):
     # load model
     self.model = pickle.load(open(model_path, 'rb'))
   def CDR_calc(self, OD_mask, OC_mask):
@@ -28,7 +28,7 @@ def CDR_calc(self, OD_mask, OC_mask):
     return VCDR, HCDR, ACDR
 
 
-  def predict(self, model, feature):
+  def predict(self, feature):
     predict = self.model.predict([feature])
     if predict[0] == 1:
       return 'Glaucoma'

Code/localization.py

@@ -6,13 +6,13 @@
 class OD_localization():
   def __init__(self):
     # loading template
-    image_R = cv2.imread('Code/Templates/ROItemplateRed.png', 0)
-    image_G = cv2.imread('Code/Templates/ROItemplateGreen.png', 0)
-    image_B = cv2.imread('Code/Templates/ROItemplateBlue.png', 0)
+    image_R = cv2.imread('Templates/ROItemplateRed.png', 0)
+    image_G = cv2.imread('Templates/ROItemplateGreen.png', 0)
+    image_B = cv2.imread('Templates/ROItemplateBlue.png', 0)
     self.image_templates = [image_R, image_G, image_B]
 
   def preprocessing(self, src, std_size):
-    return resize(src, output_shape=std_size, mode = 'constant')
+    return cv2.resize(src, std_size, interpolation=cv2.INTER_CUBIC)
 
   def extract_BR_map(self, src, mask, numSegments=50, sigma=10):
     """
@@ -121,7 +121,7 @@ def image_template_matching(self, clahe_retinal_img, template, mask):
     return NCC_maps
 
 
-  def predict(self, src, coeff_args, test_on=False):
+  def locate(self, src, coeff_args, test_on=False):
     """
       This is the main function for localization of Optic Disc
       
@@ -142,7 +142,7 @@ def predict(self, src, coeff_args, test_on=False):
       test_on is True.
     """
 
-      
+    R_COEFF, G_COEFF, B_COEFF, BR_COEFF = coeff_args
     h, w = src.shape[:2]
 
     # Implement CLAHE to the input image 
@@ -169,7 +169,7 @@ def predict(self, src, coeff_args, test_on=False):
     brightness_map = self.extract_BR_map(src, mask)
 
     # Combining localization maps 
-    combined_map = red_NCC * r_coeff + green_NCC*g_coeff + blue_NCC*b_coeff + brightness_map*br_coeff
+    combined_map = red_NCC * R_COEFF + green_NCC*G_COEFF + blue_NCC*B_COEFF + brightness_map*BR_COEFF
     all_maps = [combined_map, red_NCC, green_NCC, blue_NCC, brightness_map]
 
     # Extracting maximum value of NCC

Code/Inference_script.py → Code/main.py

@@ -1,22 +1,17 @@
 import cv2 
 import sys
 import time
-import os
-import math
-import pickle
 import numpy as np
-import tkinter as tk
-from skimage.transform import resize
-from skimage.segmentation import slic
 from matplotlib import pyplot as plt
-from keras import backend as K
-from tkinter import filedialog
 from localization import OD_localization
 from segmentation import optic_segmentation
 from classification import inference_glaucoma
 from supporting_function import *
 
 if __name__ == "__main__":
+    # take path from argument
+    file_path = sys.argv[1]
+
     # Load module objects
     ODFinder = OD_localization() # localization model
     segModel = optic_segmentation() # segmentation model
@@ -37,31 +32,30 @@
 
     # Localize Optic Disc
     start_loc = time.time()
-    disc_center = ODFinder(ret_img, coeff_args=coeff_args)
+    disc_center = ODFinder.locate(ret_img, coeff_args=coeff_args)
     end_loc = time.time()
-
+    print(disc_center)
     # OD and OC segmentation 
     start_seg = time.time()
     clahe = cv2.createCLAHE(clipLimit =2.0, tileGridSize=(8,8))
-    cl_img = clahe.apply(retinal_image[:, :, 1])
-    ROI, coordinate = ekstrakROI(ret_img, ROI_size, cl_img)
-    OD_pred, OC_pred = segModel.do_segmentation(ROI, coordinate, )
+    cl_img = clahe.apply(ret_img[:, :, 1]) # apply clahe on green channel 
+    ROI, coordinate = ekstrakROI(disc_center, ROI_SIZE, cl_img)
+    OD_pred, OC_pred = segModel.do_segmentation(ROI, coordinate, ret_img.shape[:2])
     end_seg = time.time()
 
-
     print('Localization time: {:.2f} s'.format(end_loc-start_loc))
     print('Segmentation time: {:.2f} s'.format(end_seg-start_seg))
 
     # Inference the feature
-    VCDR,HCDR,ACDR = CDR_calc(OD_mask, OC_mask)
+    VCDR,HCDR,ACDR = glPredictor.CDR_calc(OD_pred, OC_pred)
     feature = np.array([VCDR, ACDR]) # best prediction achieved by using only VCDR and ACDR
-    pred = glaucoma_predict(inference_model, feature)
+    pred = glPredictor.predict(feature)
 
-    print('Detection report of file ', filename)
+    print('Detection report')
     print('Prediction: {}'.format(pred))
 
     fig, ax = plt.subplots(1, 1,  figsize= (5, 5))
-    ax.imshow(retinal_image)
+    ax.imshow(ret_img)
     ax.contour(OC_pred, colors='b')
     ax.contour(OD_pred, colors='w')
     ax.grid(False)

Code/segmentation.py

@@ -1,6 +1,7 @@
 import cv2
 import numpy as np
 from skimage.transform import resize
+from keras import backend as K
 import tensorflow as tf
 
 def fscore(y_true, y_pred):
@@ -9,7 +10,7 @@ def fscore(y_true, y_pred):
     return 2*((precision*recall)/(precision+recall+K.epsilon()))
 
 class optic_segmentation():
-  def __init__(self, model_OD_path='Code/Models/model OD semantic', model_OC_path='Code/Models/model OC semantic'):
+  def __init__(self, model_OD_path='Models/model OD semantic', model_OC_path='Models/model OC semantic'):
     # Load semantic segmentation model 
     self.model_OD = tf.keras.models.load_model(model_OD_path,custom_objects={'fscore':fscore})
     self.model_OC = tf.keras.models.load_model(model_OC_path,custom_objects={'fscore':fscore})

README.md

@@ -2,20 +2,23 @@
 
 ## Intoduction:
 
-Glaucoma is the second cause of blindness. Unfortunately, Indonesian people still unaware for the importance of early glaucoma detection. Moreover, Glaucoma Detection is the laborous and subjective job for ophtalmologists. Thus, we need the automation for glaucoma detection. We use glaucoma detection based on Optic Cup and Disc segmentation. This automation allow us to detect glaucoma more faster and less subjective. Furthermore, this system could be implemented in smartphone, hence more patient could monitor their glaucoma stage in much more affordable way. For further algorithm explanation, you can read this paper code [here](http://www.joig.net/index.php?m=content&c=index&a=show&catid=78&id=299) 
+Glaucoma is the second cause of blindness. Unfortunately, Indonesian people still unaware for the importance of early glaucoma detection. Moreover, Glaucoma Detection is the laborous and subjective job for ophtalmologists. Thus, we need the automation for glaucoma detection. We use glaucoma detection based on Optic Cup and Disc segmentation. This automation allow us to detect glaucoma more faster and less subjective. Furthermore, this system could be implemented in smartphone, hence more patient could monitor their glaucoma stage in much more affordable way. For further algorithm explanation, you can read our paper [here](http://www.joig.net/index.php?m=content&c=index&a=show&catid=78&id=299).
 
-Below are the main algorithm for Glaucoma Detection used in this project: 
+The main algorithm for our Glaucoma Detection is follow: 
 1. Preprocessing
 2. Optic Disc Localization
 3. Optic Disc dan Cup Segmentation
 4. Glaucoma Feature Extraction
-5. Classification
+5. Glaucoma Prediction
+
+[Glaucoma detection algorithm](https://github.com/anasnafis77/Glaucoma-Detection-UNet/tree/main/readme_img/Alur algoritma for poster english.png)
 
 ## Usage
 Below are the procedure for using this script:
-1. Run this: `python inference_script.py` in command prompt. Make sure you are in 'Code' directory.
-2. Choose your retinal image.
-3. After a few seconds, the detection result should appear in command prompt and json file.
+1. Open your command line interface (command prompt, powershell, etc.) and go to project directory.
+2. Install all libraries in requirements.txt (pip install -r requirements.txt).   
+3. Run this: `python main.py /path/to/retinal_image.jpg` in command prompt. Make sure you are in 'Code' directory.
+4. After a few seconds, the detection result should appear in your command line interface and segmentation result would appear from matplotlib window.
 
 ## Notebooks:
 Main Notebook:
@@ -27,7 +30,7 @@ Localization Notebook:
 Segmentation Notebook:
 [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/anasnafis77/Deteksi-Glaukoma/blob/main/Notebooks/Segmentation_notebook.ipynb)
 
-Inferencing Notebook:
+Glaucoma prediction Notebook:
 [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/anasnafis77/Deteksi-Glaukoma/blob/main/Notebooks/Inferencing_notebook.ipynb)
 
 ## Reference

requirements.txt

@@ -0,0 +1,70 @@
+absl-py==1.3.0
+astunparse==1.6.3
+cachetools==5.2.0
+certifi==2022.9.24
+charset-normalizer==2.1.1
+click==8.1.3
+colorama==0.4.6
+cycler==0.11.0
+docopt==0.6.2
+Flask==2.2.2
+flatbuffers==22.10.26
+gast==0.4.0
+google-auth==2.13.0
+google-auth-oauthlib==0.4.6
+google-pasta==0.2.0
+grpcio==1.50.0
+h5py==3.7.0
+idna==3.4
+imageio==2.22.2
+importlib-metadata==5.0.0
+itsdangerous==2.1.2
+Jinja2==3.1.2
+joblib==1.2.0
+keras==2.9.0
+Keras-Preprocessing==1.1.2
+kiwisolver==1.4.4
+libclang==14.0.6
+Markdown==3.4.1
+MarkupSafe==2.1.1
+matplotlib==3.3.1
+networkx==2.8.7
+numpy==1.23.2
+oauthlib==3.2.2
+opencv-python==4.6.0.66
+opt-einsum==3.3.0
+packaging==21.3
+pandas==1.4.3
+Pillow==9.2.0
+pipreqs==0.4.11
+protobuf==3.19.6
+pyasn1==0.4.8
+pyasn1-modules==0.2.8
+pyparsing==3.0.9
+python-dateutil==2.8.2
+pytz==2022.5
+PyWavelets==1.4.1
+requests==2.28.1
+requests-oauthlib==1.3.1
+rsa==4.9
+scikit-image==0.19.3
+scikit-learn==1.1.3
+scipy==1.9.3
+six==1.16.0
+sklearn==0.0
+tensorboard==2.10.1
+tensorboard-data-server==0.6.1
+tensorboard-plugin-wit==1.8.1
+tensorflow==2.10.0
+tensorflow-estimator==2.10.0
+tensorflow-gpu==2.9.1
+tensorflow-io-gcs-filesystem==0.27.0
+termcolor==2.0.1
+threadpoolctl==3.1.0
+tifffile==2022.10.10
+typing-extensions==4.4.0
+urllib3==1.26.12
+Werkzeug==2.2.2
+wrapt==1.14.1
+yarg==0.1.9
+zipp==3.10.0