Paper Key : IRJ************655
Author: Shaik Afrin
Date Published: 13 Nov 2024
Abstract
Recognizing out handwritten digits is super important for recognizing patterns and it pops up in a bunch of places like entering data from forms handling bank checks and sorting mail older techniques often have a hard time with the differences in handwritten digits cause they can end up looking misaligned or are hard to read to tackle these problems. This paper takes a look at how to recognize handwritten digits using deep learning mainly focusing on convolutional neural networks cnns the vgg-16 design and multi-layer perceptrons mlps we dig into how well cnns perform since they use convolutional and pooling layers to learn features from images on their own which makes them really good at classifying images the vgg-16 model which is known for its deep structure and the regular use of small convolutional filters really helps with feature extraction and boosts accuracy in classification quite a bit on the flip side while mlps can act as classifiers they often need a lot of feature engineering and struggle to capture spatial relationships the way cnns do our study highlights the advantages of cnn-based models especially vgg-16 when compared to traditional mlps showing how deep learning is changing the game in recognizing handwritten digits and offering some ideas for future research in this area.