feature_extraction1.py

# -*- coding: utf-8 -*-
"""Feature_extraction1.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/19ThC4NyRtLoSvMLw2gY-KdjK42VW6DfU
"""

from google.colab import drive
drive.mount("/content/drive", force_remount=True)

import os
os.chdir('/content/drive/MyDrive/Colab Notebooks/Mask_RCNN/samples/videos/save2')

import tensorflow as tf 
from keras.applications.resnet50 import ResNet50
from keras.applications.vgg16 import VGG16
from keras.layers import Flatten, Input
from keras.models import Model
from keras.preprocessing import image
from keras.applications.imagenet_utils import preprocess_input
import numpy as np
import pandas as pd
import time
import cv2

model1 = ResNet50(weights='imagenet', pooling=max, include_top = False) 
model2 = VGG16(weights='imagenet', include_top=False)

root_dir = os.getcwd()
features=[]

start = time.time()
for i in range(10):

   path = os.path.join(root_dir, str(i))
   for filename in os.listdir(path):
     img_path = os.path.join(path,str(filename))
     img = image.load_img(img_path, target_size=(224, 224))
     img_data = image.img_to_array(img)
     img_data = np.expand_dims(img_data, axis=0)
     img_data = preprocess_input(img_data)
     vgg16_feature = model2.predict(img_data)
     features.append(vgg16_feature.squeeze())
     end = time.time()
   print('\ntime spend: ' , (end - start)/60 , ' minutes \n')

np.array(features).shape

extracted_features = np.array(features).reshape(810,-1)

df = pd.DataFrame(extracted_features)

df.to_csv('/content/drive/MyDrive/Colab Notebooks/Mask_RCNN/samples/videos/save2/extracted_features_vgg16.txt' , index=False, sep =' ')