Klasifikasi Social Distancing Berbasis Deep Learning Pada Data Video CCTV

Nadhil Q, M Thalut (2021) Klasifikasi Social Distancing Berbasis Deep Learning Pada Data Video CCTV. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 05111740000143-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (6MB) | Request a copy

Abstract

Dengan menyebarnya virus Corona atau COVID-19 di
Indonesia pada bulan maret 2020, menimbulkan berbagai
dampak pada tiap aspek kehidupan masyarakat Indonesia selama kurang lebih 1 tahun ke depannya. Dengan adanya new normal, diharapkan roda ekonomi Indonesia tetap berjalan sehingga ekonomi Indonesia bisa seperti dulu lagi sebelum virus ini masuk ke Indonesia, tidak lupa dengan tetap menerapkan protokol kesehatan seperti menggunakan masker atau menerapkan social distancing.
Tetapi pihak terkait mengalami kesulitan dalam
mengetahui tingkat kepatuhan masyarakat dalam mematuhi
protokol Social Distancing atau menjaga jarak pada new normal ini. Permasalahan ini dapat diselesaikan dengan bantuan salah satu model Deep Learning, yaitu SSD Mobilenet. Dimana SSD Mobilenet sebagai deteksi manusianya yang dibantu dengan algoritma tracking yaitu Kalman filter dan Hungarian Algorithm serta script lain yang membantu dalam pengklasifikasian Social Distancing.
Dataset yang digunakan untuk melakukan training model
adalah dataset Widerperson dan model hasil tersebut akan di uji cobakan ke dataset CCTV, yaitu Oxford Town Center Dataset (an urban street) dan Mall Dataset (an indoor mall). Dan hasil pengujian social distancing dua kelas yaitu melanggar dan tidak menghasilkan nilai Akurasi sebesar 68.2%, Presisi sebesar 82.3%, Recall sebesar 60.6%, dan F1 Score sebesar 69.8% pada Oxford Town Center Dataset dan Akurasi sebesar 78.3%, Presisi sebesar 94.5%, Recall sebesar 79.1%, dan F1 Score sebesar 86.1% pada Mall Dataset.
=====================================================================================================
With the spread of the Corona virus or COVID-19 in
Indonesia in March 2020, it caused various impacts on every aspect of Indonesian life for approximately 1 year to the community. With the new normal, it is hoped that the wheels of the Indonesian economy will continue to run so that the Indonesian economy can be like before before the virus entered Indonesia, don't forget to continue to apply health protocols such as wearing masks or implementing social distancing. However, related parties have difficulty knowing the level of community compliance in complying with the Social Distancing protocol or maintaining a distance in this new normal. This problem can be solved with the help of one of the
Deep Learning models, which is SSD Mobilenet. Where the
Mobilenet SSD as human detection is assisted by tracking
algorithms, namely the Kalman filter and Hungarian Algorithm and other scripts that help in classifying Social Distancing.
The dataset used to train the model is the Widerperson
dataset and the resulting model will be tested on the CCTV
dataset, which is Oxford Town Center Dataset (an urban
street) and Mall Dataset (an indoor mall). And the results of the two-class social distancing test are violating and not, producing value is 68.2% of accuracy, 82.3% of precision, 60.6% of recall, dan 69.8% of F1 Score from Oxford Town Center Dataset and the result for test of social distancing for two classes is 78.3% of accuracy,94.5% of precision, 79.1% of
recall, dan 86.1% of F1 Score from Mall Dataset.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Covid, Social Distancing, Deep Learning, Single Shot Detection (SSD), Mobilenet
Subjects:	R Medicine > R Medicine (General) > R858 Deep Learning T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.8 Productivity. Efficiency T Technology > TD Environmental technology. Sanitary engineering > TD890 Global Environmental Monitoring System
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	M Thalut Nadhil Q
Date Deposited:	07 Aug 2021 07:19
Last Modified:	07 Aug 2021 07:19
URI:	http://repository.its.ac.id/id/eprint/85059

Actions (login required)

View Item