Team: 15
School: Desert Academy
Area of Science: Computer Science
Proposal: Goal:
Our idea is to make an application that you can input an image file of a handwritten signature and the application will compare the input with a saved file. The output of the application will be whether or not the application thinks that the signature is valid.
Purpose:
The purpose of this project is to increase the security of anything that requires a signature.
Plan:
The first thing we need to do is collect data. We will ask everyone in our school to write there signature with varying amounts of speed and scan and label the data on our computer.
After getting the data we need to set up a neural network with the twice amount of input neurons as pixels in the data images. We plan to have 2 hidden layers the first having 32 neurons and the second hidden layer having 20 neurons. This is subject to change.
After having the feedforward network completed we are hoping to begin training our model using the back-propagation algorithm. After training it should be able to output a probability value that the images are the same..
Our final set is to make an application that can separate images of words into letters and then feed the letter images into our model which will output the corresponding letters.
Students: Jonatan Kaare-Rasmussen, Ethan Garcia
Mentor: Chris Zappe
Team Members:
Jonatan Kaare-Rasmussen
Ethan Garcia
Sponsoring Teacher: Christopher Zappe