230323_(진행중)_연구노트

import numpy as np #import matplotlib.pyplot as plt import os import math import shutil ROOT_DIR = '/home/test4611/Desktop/105_classes_pins_dataset' DEST_DIR = './Data' n = 5 #int(input("Enter Number of Targets: ")) #shutil.rmtree("./Data") import random exclude=[] target=[] count=[] no_of_images = {} if not os.path.exists("./Data"): os.mkdir("./Data") for i in range(n): personfolder=os.listdir(ROOT_DIR) for item in exclude: if item in personfolder: personfolder.remove(item) person=random.choice(personfolder) src = (ROOT_DIR+'/'+person) dest = ('./Data/'+person[5:]) destexcl = (person) exclude.append(destexcl) shutil.copytree(src,dest) for dir in os.listdir(DEST_DIR): no_of_images[dir] = len(os.listdir(os.path.join(DEST_DIR,dir))) othersno=max(no_of_images.values()) if not os.path.exists("./Data/NON-TARGET"): os.mkdir("./Data/NON-TARGET") for i in range(othersno+2): person=random.choice(personfolder) files=os.listdir(ROOT_DIR+'/'+person) face=random.choice(files) src = (ROOT_DIR+'/'+person+'/'+face) dest = ('./Data/NON-TARGET/') shutil.copy(src,dest) no_of_images = {} for dir in os.listdir(DEST_DIR): no_of_images[dir] = len(os.listdir(os.path.join(DEST_DIR,dir))) no_of_images.items() print('\n') for item in no_of_images: if item != 'NON-TARGET': target.append(item) print('Target_Name\t\tNo_of_Images') for item in target: print(item,'\t\t',no_of_images[item]) print('NON-TARGET','\t\t',no_of_images['NON-TARGET']) ### def dataFolder(p, split): if not os.path.exists("./Model_Data"): os.mkdir("./Model_Data") if not os.path.exists("./Model_Data/"+p): os.mkdir("./Model_Data/"+p) for dir in os.listdir(DEST_DIR): os.makedirs("./Model_Data/"+p+"/"+dir) for img in np.random.choice(a=os.listdir(os.path.join(DEST_DIR,dir)), size = (math.floor(split*no_of_images[dir])-5),replace=False): O = os.path.join(DEST_DIR,dir,img) D = os.path.join("./Model_Data/"+p,dir) shutil.copy(O,D) else: print(f"{p} Exists") dataFolder("Train",0.7) dataFolder("Val",0.3) dataFolder("Test",0.5) #### from keras.preprocessing.image import ImageDataGenerator def preprocessingTrain(path): image_data = ImageDataGenerator(featurewise_center=True, rotation_range=0.4, width_shift_range=0.3, zoom_range=0.2, shear_range=0.2, rescale=1./255, horizontal_flip= True) image = image_data.flow_from_directory(directory= path, target_size=(256,256), batch_size=2, class_mode='categorical') return image def preprocessingVal(path): image_data = ImageDataGenerator(rescale=1./255) image = image_data.flow_from_directory(directory= path, target_size=(256,256), batch_size=2, class_mode='categorical') return image ### train_path="./Model_Data/Train" train_data = preprocessingTrain(train_path) val_path="./Model_Data/Val" val_data = preprocessingVal(val_path) test_path="./Model_Data/Test" test_data = preprocessingVal(test_path) from keras.models import Model from keras.layers import Flatten, Dense import keras.losses from keras.applications.inception_resnet_v2 import InceptionResNetV2 as IncRes ### base_model = IncRes(input_shape=(256,256,3),weights='imagenet',include_top=False) for layer in base_model.layers: layer.trainable=False X=Flatten()(base_model.output) X=Dense(units=n+1, activation='softmax')(X) model_IncRes = Model(base_model.input, X) model_IncRes.compile(optimizer='adam',loss=keras.losses.categorical_crossentropy,metrics=['accuracy']) model_IncRes.summary() #model_IncRes.summary() ###

import h5py import numpy as np #import matplotlib.pyplot as plt #%matplotlib inline # # used to supress display of warnings import warnings #from sklearn.metrics import precision_recall_curve,accuracy_score,f1_score,precision_score,recall_score # suppress display of warnings warnings.filterwarnings('ignore') import os source_dir=os.path.join('/home/test4611/Desktop/105_classes_pins_dataset') class IdentityMetadata(): def __init__(self, base, name, file): self.base = base # identity name self.name = name # image file name self.file = file def __repr__(self): return self.image_path() def image_path(self): return os.path.join(self.base, self.name, self.file) def load_metadata(path): metadata = [] for i in os.listdir(path): for f in os.listdir(os.path.join(path, i)): # Check file extension. Allow only jpg/jpeg' files. ext = os.path.splitext(f)[1] if ext == '.jpg' or ext == '.jpeg': metadata.append(IdentityMetadata(path, i, f)) return np.array(metadata) # metadata = load_metadata('images') metadata = load_metadata(source_dir) from tensorflow.keras.models import Sequential from tensorflow.keras.layers import ZeroPadding2D, Convolution2D, MaxPooling2D, Dropout, Flatten, Activation def vgg_face(): model = Sequential() model.add(ZeroPadding2D((1,1),input_shape=(224,224, 3))) model.add(Convolution2D(64, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(64, (3, 3), activation='relu')) model.add(MaxPooling2D((2,2), strides=(2,2))) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(128, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(128, (3, 3), activation='relu')) model.add(MaxPooling2D((2,2), strides=(2,2))) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(256, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(256, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(256, (3, 3), activation='relu')) model.add(MaxPooling2D((2,2), strides=(2,2))) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(512, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(512, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(512, (3, 3), activation='relu')) model.add(MaxPooling2D((2,2), strides=(2,2))) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(512, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(512, (3, 3), activation='relu')) model.add(ZeroPadding2D((1,1))) model.add(Convolution2D(512, (3, 3), activation='relu')) model.add(MaxPooling2D((2,2), strides=(2,2))) model.add(Convolution2D(4096, (7, 7), activation='relu')) model.add(Dropout(0.5)) model.add(Convolution2D(4096, (1, 1), activation='relu')) model.add(Dropout(0.5)) model.add(Convolution2D(2622, (1, 1))) model.add(Flatten()) model.add(Activation('softmax')) return model model = vgg_face() model.load_weights('/home/test4611/Desktop/asd/vgg_face_weights.h5')