Image_classification_of_CIFAR10_dataset

The CIFAR-10 dataset contains 60,000 color images of 32 x 32 pixels in 3 channels divided into 10 classes. Each class contains 6,000 images. The training set contains 50,000 images, while the test sets provides 10,000 images.This is a classification problem with 10 classes(muti-label classification). We can take a view on this image for more comprehension of the dataset.

More about the dataset

More details can be found in the below link. https://www.cs.toronto.edu/~kriz/cifar.html

Data lables using Categorical Encoding

The output image class is categorically encoded with the values randing from 0 to 9. The classes are:

Label	Description
0	airplane
1	automobile
2	bird
3	cat
4	deer
5	dog
6	frog
7	horse
8	ship
9	truck

Models involved

• The classification is done using ANN.
• The same classification is done using CNN as well.

• The results gathered by using an Artificial Neural Network is as below.
• Details of the neural network
  • Activation Functions: Sigmoid, ReLU
  • Loss: Categorical Cross Entropy
  • Optimiser: Adam
  • Metric: Accuracy
  • Epochs: 5

category	precision	recall	f1-score	support
0	0.48	0.55	0.51	1000
1	0.63	0.55	0.59	1000
2	0.41	0.20	0.27	1000
3	0.32	0.27	0.29	1000
4	0.49	0.26	0.34	1000
5	0.37	0.42	0.40	1000
6	0.43	0.60	0.51	1000
7	0.44	0.61	0.51	1000
8	0.57	0.62	0.60	1000
9	0.51	0.56	0.54	1000
accuracy			0.47	10000
macro avg	0.47	0.47	0.45	10000
weighted avg	0.47	0.47	0.45	10000

• The results gathered by using a Convolutional Neural Network is as below.
• Details of the neural network
  • Activation Functions: ReLU, Softmax
  • Loss: Categorical Cross Entropy
  • Optimiser: Adam
  • Metric: Accuracy
  • Epochs: 5

category	precision	recall	f1-score	support
0	0.73	0.77	0.75	1000
1	0.84	0.78	0.81	1000
2	0.49	0.66	0.56	1000
3	0.51	0.52	0.51	1000
4	0.73	0.49	0.59	1000
5	0.61	0.53	0.57	1000
6	0.85	0.70	0.77	1000
7	0.66	0.81	0.72	1000
8	0.79	0.83	0.81	1000
9	0.76	0.80	0.78	1000
accuracy			0.69	10000
macro avg	0.70	0.69	0.69	10000
weighted avg	0.70	0.69	0.69	10000

Results

As we can see, for the same number of epochs there is a huge change in Artificial and Convolutional Neural network. The performance in the metrics like precision, recall, f1-score, etc. has increased for all the 10 categories.