import sys, os
sys.path.append(os.path.abspath('..\\..\\ezmodel'))
from ezmodel.ezmodel import ezmodel
from ezmodel.ezdata import ezdata
from ezmodel.eztrainer import eztrainer
# [EZDATA]
parameters = {
"name": "Skin to predict",
"path": "C:\\Users\\daian\\Desktop\\DATA\\Skin\\skin_to_predict.npz",
"resize": (128, 128)
}
data_to_predict = ezdata(parameters)
# [EZNETWORK]
ez_trainer = eztrainer(network="imagenet.mobilenet")
# [EZMODEL]
ez_model = ezmodel()
ez_model.assign(data_to_predict, ez_trainer)
ez_model.predict(data_to_predict)
# ---------------------------- [EZ Data] -----------------------------------
# Table Classification from a csv file:
# - One subdirectory by Class
parameters = {
"name": "Iris",
"type": "classification",
"format": "table",
"from": "file",
"path": "C:\\Users\\daian\\Desktop\\DATA\\Iris\\iris.csv",
"table_delimiter": ",",
"table_target_column": ["species"],
"table_target_column_type": "string"
}
ez_data = ezdata(parameters)
ez_data.gen_test(size=0.2)
ez_data.preprocess(X="standard",
y="categorical") #on crée les scaler dans ez_data aussi
# ---------------------------- [EZ Data Save] -------------------------------
ez_data.save("iris_dataset")
# ---------------------------- [EZ Data Load] -------------------------------
ez_data2 = ezdata(load="iris_dataset")
print(ez_data2.params["name"])
from ezmodel.eztrainer import eztrainer
# ---------------------------- [EZ Data] -----------------------------------
# Load from a npz file containing data
# NPZ file should contains 3 keys:
# X key containing data
# y key containing label
# synsets key in case of classification model with synset generated
parameters = {
"name" : "Skin",
"path" : "C:\\Users\\daian\\Desktop\\DATA\\Skin\\skin.npz",
"X.key" : "images",
"y.key" : "labels"
}
ez_data = ezdata(parameters)
ez_data.gen_test(0.2)
#ez_data.preprocess(X="mobilenet",y="categorical")
ez_data.preprocess(X="vgg19",y="categorical")
# -------------------------- [EZ Trainer] ------------------------------------
from keras.layers import Dense,Dropout
ez_trainer = eztrainer()
ez_trainer.gen_trainval(ez_data,size=0.2)
# -- Keras network --
import sys, os
sys.path.append(os.path.abspath('..\\..\\ezmodel'))
from ezmodel.ezmodel import ezmodel
from ezmodel.ezdata import ezdata
from ezmodel.eztrainer import eztrainer
ezd = ezdata()
ezd.X = ezd.image_from_url(
url=
"https://www.almanac.com/sites/default/files/styles/primary_image_in_article/public/image_nodes/strawberries-1.jpg",
img_size=(128, 128))
ezd.y = None
import pickle
from urllib.request import urlopen
ezd.synsets = pickle.load(
urlopen(
'https://gist.githubusercontent.com/yrevar/6135f1bd8dcf2e0cc683/raw/d133d61a09d7e5a3b36b8c111a8dd5c4b5d560ee/imagenet1000_clsid_to_human.pkl'
))
ezt = eztrainer()
ezt.network = ezt.PretrainedNetwork("mobilenet", ezd.X.shape[1:])
ezt.network.summary()
ezd.preprocess(X="mobilenet", y=None)
import numpy as np
p = ezt.network.predict(ezd.X)
top = 5
ind_five = (-p).argsort()[-3:][::-1][0][0:top]