import numpy as np
import argparse
import os
import tensorflow as tf
import glob
import matplotlib.pyplot as plt
import cv2
import random

ap = argparse.ArgumentParser()
ap.add_argument("-d", "--dataset", type=str, default = "/home/imagda/sims-data/malaria",  help="path dataset of input images")
ap.add_argument("-m", "--model", type=str, default = "orig/saved_model.model", help="path to trained model")
ap.add_argument("-p", "--plot", type=str, default="plot.png", help="path to output loss/accuracy plot")
args = vars(ap.parse_args([]))

1. Laden des Modelles

tf.keras.models.load_model(path.h5)
if model was saved as h5 no additional information to the model architecture, weights etc is needed

model = tf.keras.models.load_model(args["model"])

2. Datenvorverarbeitung gemäß den Trainingsdaten

tot_test_paths =  glob.glob(os.path.sep.join([args["dataset"], "test", "*", "*"]))
random.shuffle(tot_test_paths)
print(len(tot_test_paths))

2756

test_images = tot_test_paths[:10]
# initialize a list where to put the new images with label
res_imgs = []

3. Modellvorhersage und Ergebnisdarstellung

for i, image in enumerate (tot_test_paths[:5]):
    # load input image
    orig = cv2.imread(image)
    # convert from cv2 default BGR --> RGB
    img = cv2.cvtColor(orig, cv2.COLOR_BGR2RGB)
    # resize image to size accepted by model
    img = cv2.resize(img, (64,64))
    # rescale image between 0 - 1 floats (previously was at 0- 255 integer)
    img = img.astype("float")/255.
    # converts a pil image to array
    img = tf.keras.preprocessing.image.img_to_array(img)
    # img array will get an additional dimension fro batch  img(batch=1, R, G, B, 3)
    img = np.expand_dims(img, axis=0)
    # using preterained model make a prediction
    res = model.predict(img)
    # max prediction index
    pred = res.argmax(axis=1)[0]
    label = "Parasitized" if pred == 0 else "Uninfected"
    color = (0,0, 255) if pred == 0 else (0,255, 0)
    orig = cv2.resize(orig, (128,128))
    cv2.putText(orig, label, (3, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)
    res_imgs.append(orig)

plt.imshow(res_imgs[0])

<matplotlib.image.AxesImage at 0x7fa468d20d50>

References

Adrian Rosebrock, OpenCV Face Recognition, PyImageSearch, https://www.pyimagesearch.com/, accessed on 3 January, 2021> www:https://www.pyimagesearch.com/2018/12/10/keras-save-and-load-your-deep-learning-models/