Weather Classifcation

Credit: AITS Cainvas Community

Photo by Sergey Galtsev on Dribbble

Image tagging helps in selecting images based on content, especially useful in search engines and other similar applications. Here, we tag images based on the weather of the scene. There are two classes - cloudy, sunny.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix
from tensorflow.keras import layers, optimizers, models, preprocessing, losses, callbacks
import os
import random
from PIL import Image
import tensorflow as tf
import tensorflow.keras

Dataset

"Two-class Weather Classification" Cewu Lu, Di Lin, Jiaya Jia, Chi-Keung Tang IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014

On Kaggle by Paula

!wget https://cainvas-static.s3.amazonaws.com/media/user_data/cainvas-admin/weather.zip
 
!unzip -qo weather.zip

!rm weather.zip

--2021-09-07 06:46:34--  https://cainvas-static.s3.amazonaws.com/media/user_data/cainvas-admin/weather.zip
Resolving cainvas-static.s3.amazonaws.com (cainvas-static.s3.amazonaws.com)... 52.219.64.104
Connecting to cainvas-static.s3.amazonaws.com (cainvas-static.s3.amazonaws.com)|52.219.64.104|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 116150379 (111M) [application/zip]
Saving to: ‘weather.zip’

weather.zip         100%[===================>] 110.77M   108MB/s    in 1.0s    

2021-09-07 06:46:35 (108 MB/s) - ‘weather.zip’ saved [116150379/116150379]

# Loading the dataset

path = 'weather/'
input_shape = (256, 256, 3)    # default input shape while loading the images

batch = 64

# The train and test datasets
print("Train dataset")
train_ds = preprocessing.image_dataset_from_directory(path+'train', batch_size=batch, label_mode='binary')

print("Test dataset")
test_ds = preprocessing.image_dataset_from_directory(path+'test', batch_size=batch, label_mode='binary')

Train dataset
Found 10000 files belonging to 2 classes.
Test dataset
Found 253 files belonging to 2 classes.

# How many samples in each class

for t in ['train', 'test']:
    print('\n', t.upper())
    for x in os.listdir(path + t):
        print(x, ' - ', len(os.listdir(path + t + '/' + x)))

 TRAIN
sunny  -  5000
cloudy  -  5000

 TEST
sunny  -  153
cloudy  -  100

The train set is balanced while the test set is imbalanced. A confusion matrix can help in finding the accuracies.

# Looking into the class labels

class_names = train_ds.class_names

print("Train class names: ", train_ds.class_names)
print("Test class names: ", test_ds.class_names)

Train class names:  ['cloudy', 'sunny']
Test class names:  ['cloudy', 'sunny']

Visualization

num_samples = 4    # the number of samples to be displayed in each class

for x in class_names:
    plt.figure(figsize=(20, 20))

    filenames = os.listdir(path + 'train/' + x)

    for i in range(num_samples):
        ax = plt.subplot(1, num_samples, i + 1)
        img = Image.open(path +'train/' + x + '/' + filenames[i])
        plt.imshow(img)
        plt.title(x)
        plt.axis("off")