GO MUSEUM
Here is where your presentation begins
Team:
Karim Yasser, Ahmed Amr
Amr Salama & Loay Yehia
Supervised By:
Dr. Fatma Helmy
Supervising Assistant:
Eng. Samira Raafat
Date: 3/8/2022
TABLE OF CONTENTS
01
02
03
04
Introduction & Objectives
Problem Definition
System Overview
Methodology
Diagrams
Dataset Description
User interface
Results
What's Next & Demo
05
06
07
08
09
Main Objectives
The goal is to create a user-friendly mobile application that will allow tourists to visit historical sites without having to hire a tour guide, facilitate the tourist tour by guiding them to have the best experience
GO MUSEUM:
The main problem focused on is monuments detection. Firstly, the aim of Go Museum is the recognition of statues. However, the main problem Go Museum is trying to solve is addressing occluded, disoriented, and distorted monuments
PROBLEM STATMENT
System Overview
Architecture Diagram
Package diagram
Sequence diagram
Class Diagram
Decorator Design Pattern
Context Diagram
3,149
Original + Augmented + online dataset collection
Original images
Online images
1,724
From
to
Methodology
DATA Preprocessing
GAN MODEL
Outpainting
ResNet50
TRAINING PROBLEMS
Training with random mask on 2 classes
center mask training with 2 classes
training with random mask on the whole dataset
Achieved Results
Random Crop
Center Crop
Original image
ResNet50 Model
Acc : 96.3%
Predict the image Class
test
train
25%
75%
Overall Algorithm
Semi-Supervised deep learning(TripGAN)
GAN (upsamling & downsampling) layers
Generate image with Random/Center Crop
Generator
Classify if the generated image is real or fake
Discriminator
Classify the image class based on pretrained model
Classifier
What’s Next ?
A PICTURE IS WORTH A THOUSAND WORDS
THANK YOU