Pengembangan handwriting recognition dengan menggunakan algoritma CNN

Gamendra Chasbulloh, Muhammad (2018) Pengembangan handwriting recognition dengan menggunakan algoritma CNN. Project Report. studentstaff5, Perpustakaan Institut Teknologi Telkom Purwokerto. (Unpublished)

	Text Cover.pdf Download (169kB)
Preview	Text Abstrak.pdf Download (11kB) | Preview
Preview	Text Abstract.pdf Download (11kB) | Preview
Preview	Text BAB I.pdf Download (67kB) | Preview
	Text BAB II.pdf Restricted to Registered users only Download (28kB) | Request a copy
	Text BAB III.pdf Restricted to Registered users only Download (421kB) | Request a copy
	Text BAB IV.pdf Restricted to Registered users only Download (12kB) | Request a copy
Preview	Text Daftar Pustaka.pdf Download (67kB) | Preview
Preview	Text Lampiran.pdf Download (42kB) | Preview

Abstract

In the millennial era, the development of technology is so rapid that it can be proven by many developments carried out by the government in the Smart City sector. For example, Jakarta government is streaming Virtual Reality. In the development of smart cities, government usually cooperates with private companies such as Solusi 247. The company is engaged in BIG data field. One application made by the company is the Chanthel application, but the application has a disadvantage that is not able to do indexing and content. To overcome this problem, a system of Handwriting Recognition is made which aims to help the process of indexing and content. In making Handwriting Recognition requires a dataset, the data is obtained through questionnaire sharing and from the MNIST website. The data processing process uses the CNN (Convolution Neural Network) algorithm. In the execution of the Handwriting Recognition project, it is divided into 4 teams, namely the data processing team, the model making team, the team back end, and the fire team rest. The author himself gets a part making models. The steps taken by the author in making the model are understanding the concept, mengexplore library, data labeling process, creating a model using the CNN (Convolution Neural Network) algorithm. The results obtained from making the model with a number of 280,372 datasets and epoch 50 get an accuracy rate of 78.96% and loss of 0.5894. Keyword : Handwriting Recognition, CNN algorithm, Model

Actions (login required)

View Item