Add files via upload

Makefile

@@ -0,0 +1,56 @@
+# Makefile
+
+# Shell
+SHELL := /bin/bash
+#Project
+PROJECT = "image_pro"
+#Version
+RELEASE = "0.0.1"
+# Path to source directory
+PROJECTPATH := .
+# OS machine
+OS_TYPE = 'uname -a'
+# CPP EXTENSIONS
+CPP_EXT = ".cpp"
+
+# SETUP
+SETUPFILE = "setup.py"
+
+# VENV
+
+_virtualenv:
+	# create virtual env
+	virtualenv _virtualenv
+	_virtualenv/bin/pip install --upgrade pip
+	_virtualenv/bin/pip install --upgrade setuptools
+
+_use_env:
+	# use virtual env
+	source _virtualenv/bin/activate
+
+##### PYTHON
+
+prebuild:
+	@echo "Preparing build"
+	@echo "Installing requirements"
+	pip install -r requirements.txt
+
+setup:
+	echo "Setup..."
+	sudo python setup.py install
+
+test:
+	echo "Test"
+	python setup.py test
+
+clean:
+	rm -f MANIFEST
+	rm -rf build dist
+
+deactivate_env:
+	deactivate
+
+
+all: prebuild setup
+
+.PHONY: clean

image_pro/__init__.py

@@ -0,0 +1 @@
+#NOTE

image_pro/parameters.py

@@ -0,0 +1,7 @@
+UR_DATA_PATH = "./train/*/*.jpg"
+UR_PRED_PATH = "./test1/test1/*.jpg"
+UR_OPTION = "gray"
+UR_TEST_SIZE = 1000
+T_HEIGHT = 256
+T_WIDTH = 256
+T_Channnels = 1

image_pro/tfRecord/Tfrecord.py

@@ -0,0 +1,171 @@
+import os
+import cv2
+import sys
+import glob
+import numpy as np
+import tensorflow as tf
+
+from random import shuffle, seed
+from image_pro.utils_data.contours.contours import draw_all_contours
+
+
+# This function returns the different paths and corresponding labels
+def shuffling_data(data_path, labels):
+    c = list(zip(data_path, labels))
+    shuffle(c)
+    #NOTE: data and labeld are tuples. Care about immutability 
+    data, labeled = zip(*c) #NOTE:ith element in data correpand to ith label elmt
+
+    return data, labeled
+
+def read_paths_treat(ur_path, test_size):
+    pathname = os.path.dirname(ur_path)
+    #Reading the path of each data(ex: image) and then extracting the labels from path
+    data_path = glob.glob(ur_path)
+    '''labels = [int(i.split('.')[0][-1]) for i in data_path]'''#xray
+    #NOTE: labels and labels_num: modify split method to get labels// conditions "cat" or "dog"
+    labels = [i.split('\\')[-1] for i in data_path]  #Cat vs Dog
+    labels_num = [0 if "cat" in i else 1 for i in labels] #Cat vs Dog
+    len_d = len(labels_num)
+    k = int(len_d/2)
+    a = int((len_d-test_size)/2)
+    b = int(len_d-(test_size/2))
+    data_path_test, labels_test = data_path[a : k], labels_num[a:k]
+    data_path_test = data_path_test + data_path[b:]
+    labels_test = labels_test + labels_num[b:]
+    data_path_train,labels_train = (data_path[0:a]), (labels_num[0:a])
+    data_path_train = data_path_train + data_path[k:b]
+    labels_train = labels_train + labels_num[k:b]
+    data_path_test, labels_test = shuffling_data(data_path_test, labels_test)
+    data_path_train, labels_train = shuffling_data(data_path_train, labels_train)
+    return data_path_test, labels_test,data_path_train, labels_train
+
+
+def split_data(training_prop,test_prop, data, labels, dev_prop=0):
+    ''' 
+    - training_prop is required .numerical float < 1
+    - dev_prop is optional. Numerical float < 0.3
+    - test_prop is required. Numerical float < 0.3
+    - data and labels are lists 
+    '''
+    #NOTE: data is a list of paths of each image
+
+    training_data = data[0:int(training_prop*len(data))]
+    training_labels = labels[0:int(training_prop*len(labels))]
+
+    if dev_prop != 0:
+        dev_data = data[int(training_prop*len(data)):int((1-dev_prop)*len(data))]
+        dev_labels = labels[int(training_prop*len(labels)):int((1-dev_prop)*len(labels))]
+        test_data = data[int((1-dev_prop)*len(data)):]
+        test_labels = labels[int((1-dev_prop)*len(labels)):]
+        return training_data,training_labels, dev_data, dev_labels, test_data, test_labels
+
+    else:
+        test_data = data[int((1-test_prop)*len(data)):]
+        test_labels = labels[int((1-test_prop)*len(labels)):]
+        return training_data, training_labels, test_data, test_labels
+
+def load(ur_path, option):
+
+    if option =="gray":
+        img = cv2.imread(ur_path,cv2.IMREAD_GRAYSCALE)
+    elif option =="unchanged":
+        img = cv2.imread(ur_path, cv2.IMREAD_UNCHANGED)
+    else:
+        img = cv2.imread(ur_path,cv2.IMREAD_COLOR)
+    return img
+
+
+# Load image 
+
+def load_image(data, s_width, s_height, option):
+
+    #cv2 load data from data (value of one path) 
+    img_load = load(data, UR_OPTION)
+    '''img_after = cv2.cvtColor(img_load, cv2.COLOR_BGR2GRAY)'''
+    img= cv2.resize(img_cont, dsize=(s_width, s_height), interpolation=cv2.INTER_CUBIC)
+    img_f = img.tostring()
+
+    return img_f
+
+
+def _int64_feature(value):
+
+    return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))
+
+
+
+def _bytes_feature(value):
+
+    return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
+
+
+
+def get_tfrecord_file(ur_path, test_size, s_width, s_height,option, *x_filename,):
+
+    #NOTE:data is a tuple (list) of paths of each image
+    #NOTE: data can come from the dev-set or test-set
+    #NOTE: Same for labeled
+
+    test_img, test_label, train_img, train_label = read_paths_treat(ur_path, test_size)
+    file_path = x_filename[0]
+    #Open a TFRecordWriter
+
+    writer = tf.python_io.TFRecordWriter(file_path)
+    for i in range(len(train_img)):
+        new_img = load_image(train_img[i], s_width, s_height,option)
+        new_label = train_label[i]
+        new_feature = { 'height': _int64_feature(s_height),
+                        'width' : _int64_feature(s_width),
+                        'label': _int64_feature(new_label),
+                        'image': _bytes_feature(tf.compat.as_bytes(new_img))
+        }
+
+        #NOTE: PLease refer to this url for defenirtion: https://github.com/tensorflow/tensorflow/blob/r1.7/tensorflow/core/example/example.proto
+        example = tf.train.Example(features=tf.train.Features(feature=new_feature))
+        writer.write(example.SerializeToString())
+
+    writer.close()
+    sys.stdout.flush()
+    file_path = x_filename[1]
+    writer = tf.python_io.TFRecordWriter(file_path)
+
+    for j in range(len(test_img)):
+
+        new_img = load_image(test_img[j], s_width, s_height,option)
+        new_label = test_label[j]
+        new_feature = { 'height': _int64_feature(s_height),
+                        'width' : _int64_feature(s_width),
+                        'label': _int64_feature(new_label),
+                        'image': _bytes_feature(tf.compat.as_bytes(new_img))
+        }
+
+        #NOTE: PLease refer to this url for def: https://github.com/tensorflow/tensorflow/blob/r1.7/tensorflow/core/example/example.proto
+        example = tf.train.Example(features=tf.train.Features(feature=new_feature))
+        writer.write(example.SerializeToString())
+
+    writer.close()
+    sys.stdout.flush()
+    return 0
+
+
+def get_tfrecord_pred(ur_path, s_width, s_height,option, x_filename):
+    pathname = os.path.dirname(ur_path)
+    data_path = glob.glob(ur_path)
+    file_path = x_filename
+    writer = tf.python_io.TFRecordWriter(file_path)
+    for i in range(len(data_path)):
+        new_img = load_image(data_path[i], s_width, s_height,option)
+        new_feature = { 'height': _int64_feature(s_height),
+                        'width' : _int64_feature(s_width),
+                        'image': _bytes_feature(tf.compat.as_bytes(new_img))
+                        }
+        example = tf.train.Example(features=tf.train.Features(feature=new_feature))
+        writer.write(example.SerializeToString())
+    writer.close()
+    sys.stdout.flush()
+    return 0
+
+
+'''get_tfrecord_pred(UR_PRED_PATH, 256, 256,UR_OPTION, "tfpred.tfrecords")'''
+get_tfrecord_file(UR_DATA_PATH, UR_TEST_SIZE, 128, 128,UR_OPTION, "tftrain.tfrecords", "tftest.tfrecords")

image_pro/tfRecord/__init__.py

@@ -0,0 +1 @@
+#NOTE

image_pro/tfRecord/decode.py

@@ -0,0 +1,117 @@
+import os
+import numpy as np
+import tensorflow as tf
+import matplotlib.pyplot as plt
+
+#NOTE: refer to this url: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/examples/how_tos/reading_data/fully_connected_reader.py
+
+def tfrec_data_input_fn(filenames, num_epochs=1, batch_size=16, shuffle=False):
+
+    def _input_fn():
+        def _parse_record(tf_record):
+            features = {
+                'image': tf.FixedLenFeature([], dtype=tf.string),
+                'label': tf.FixedLenFeature([], dtype=tf.int64)
+            }
+            record = tf.parse_single_example(tf_record, features)
+
+            image_raw = tf.decode_raw(record['image'], tf.uint8)
+            label = tf.one_hot(tf.cast(record['label'], tf.int32), depth=2)
+
+            def _normalize(image_x):
+                """Convert `image` from [0, 255] -> [-0.5, 0.5] floats."""
+                image = tf.cast(image_x, tf.float32) * (1. / 255)-0.5
+                return image
+
+            image_raw = _normalize(image_raw)
+            '''image_raw = tf.reshape(image_raw, (32,32,3))'''#For vizualizing 
+            return { 'image': image_raw}, label
+
+        # For TF dataset blog post, see https://developers.googleblog.com/2017/09/introducing-tensorflow-datasets.html
+        dataset = tf.data.TFRecordDataset(filenames)
+        dataset = dataset.map(_parse_record)
+
+        dataset = dataset.repeat(num_epochs)
+        dataset = dataset.batch(batch_size)
+
+        iterator = dataset.make_initializable_iterator()
+        '''features, labels = iterator.get_next()'''
+
+        return iterator
+
+    return _input_fn
+
+
+def read_file(x_filename, x_capacity=800):
+    record_iterator = tf.python_io.tf_record_iterator(path=x_filename)
+
+    example = tf.train.Example()
+    for str_rec in record_iterator:
+        example.ParseFromString(str_rec)
+        height = int(example.features.feature['height']
+                                 .int64_list
+                                 .value[0])
+
+        width = int(example.features.feature['width']
+                                .int64_list
+                                .value[0])
+
+
+        img_string = (example.features.feature['image']
+                                  .bytes_list
+                                  .value[0])
+
+        label = (example.features.feature['label'].int64_list.value[0])
+
+
+    return 0
+
+
+def tfrec_data_catvdog(filenames, num_epochs=1, batch_size=16):
+
+    def _input_fn():
+        def _parse_record(tf_record):
+            features = {
+                'image': tf.FixedLenFeature([], dtype=tf.string),
+            }
+            record = tf.parse_single_example(tf_record, features)
+
+            image_raw = tf.decode_raw(record['image'], tf.uint8)
+
+            def _normalize(image):
+                """Convert `image` from [0, 255] -> [-0.5, 0.5] floats."""
+                image = tf.cast(image, tf.float32) * (1. / 255)-0.5
+                return image
+
+            image_n = _normalize(image_raw)
+            return { 'image': image_n }
+
+        # For TF dataset blog post, see https://developers.googleblog.com/2017/09/introducing-tensorflow-datasets.html
+        dataset = tf.data.TFRecordDataset(filenames)
+        dataset = dataset.map(_parse_record)
+
+        '''dataset = dataset.repeat(num_epochs)'''
+        '''dataset = dataset.batch(batch_size)'''
+
+        iterator = dataset.make_one_shot_iterator()
+        features= iterator.get_next()
+
+        return features
+
+    return _input_fn
+
+
+'''read_file("./tftest.tfrecords")
+
+tfrec_dev_input_fn = tfrec_data_input_fn(["tftest.tfrecords"])
+features, labels = tfrec_dev_input_fn()
+
+with tf.Session() as sess:
+    for step in range(2):
+        img, label = sess.run([features['image'], labels])
+        img = np.reshape(img, (16,64,64,3))
+
+        for i in range(16):
+            plt.imshow(img[i])
+            plt.show()'''
+

image_pro/utils_data/__init__.py

@@ -0,0 +1 @@
+#NOTE

image_pro/utils_data/colors/__init__.py

@@ -0,0 +1 @@
+#NOTE