Skip to main content

Table 2 Summary of newly created datasets

From: Ensemble of deep learning and machine learning approach for classification of handwritten Hindi numerals

Dataset

Originated from

Feature vector size

Size of dataset

Input

Pixel values of numeral images

1032

20,000

D1

VGG-16Net

20

20,000

D2

VGG-19Net

20

20,000

D3

ResNet-50

20

20,000

D4

Inception-v3

20

20,000

D5

Fusion of features related to D1, D2, D3, & D4

80

20,000