PS- In the given dataset classify the blurr and non blurr images.
Use the conda environment or Google Colab to see the file .ipynb and .py file attached::
Image_classification_blurr_nonblurr.ipynb
Image_classification_blurr_nonblurr.pyimport numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.applications import MobileNetV2
from tensorflow.keras.layers import AveragePooling2D
from tensorflow.keras.layers import Dropout
from tensorflow.keras.layers import Flatten
from tensorflow.keras.layers import Dense
from tensorflow.keras.layers import Input
from tensorflow.keras.models import Model
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.applications.mobilenet_v2 import preprocess_input
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.preprocessing.image import load_img
from tensorflow.keras.utils import to_categorical
from sklearn.preprocessing import LabelBinarizer
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
from imutils import paths
import matplotlib.pyplot as plt
import numpy as np
import argparse
import osSplitted the data into test set and train set
Processed the images into (224,224) pixel size
Augmentated the test data
Its a light weight model
used various models but accuracy was good in this model
epochs - 10
training accuracy- 83.23
validation accuracy- 78.91
Auc - 0.82
train loss v/s val loss
train accuracy v/s val accuracy
Kd_model.h5 (file included)
19/20 images gave the correct result
leading to a good real time accuracy