Omar Sufer (2123959) Omar Sufer

Cephalopod: a sketch-based searching tool

Project Abstract

– I find AI interesting, especially in trying to solve/improve niche problems. As a person who remembers things visually, I was motivated to create a tool that can showcase, that sketch based search tools could work- I aim to show that a sketch based system using a deep learning model can work- Using a basic website to collect data and then creating deep learning model with CNNs I was able to approach this issue and find a solution – *No major findings yet* Ive found that it could work as a niche solution tackling a specific section , e.g. in my project I focus on only searching through cover arts for music albums. Ive collected a unique dataset of 1000 sketches for 5 albums that the deep learning model can use to predict.- Conclusion, *nothing yet*

Keywords: Deep Learning searching tool, Deep Neural Networks,

 

 Conference Details

 

Session: Poster Session B at Poster Stand 106

Location: Sir Stanley Clarke Auditorium at Wednesday 8th 09:00 – 12:30

Markers: Matt Roach, Xianghua Xie

Course: BSc Computer Science, 3rd Year

Future Plans: I’m looking for work