Računalni model strojnog učenja za sortiranje otpada

undergraduate thesis

• Title (english): Machine learning computational model for waste sorting
• Author: Dora Baronica
• Mentor: Tomislav Stipančić (mentor)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2023-09-28, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Mechanical Engineering; Precision Mechanical Engineering

Abstract

Globalni problem upravljanja smećem zahtijeva kreativne pristupe pojednostavljenju postupaka sortiranja. Moderni tokovi otpada predstavljaju jedinstvene izazove za konvencionalne pristupe, zbog čega je nužno korištenje najsuvremenijih tehnologija. Ovim radom cilj je prikazati mogućnosti korištenja potencijala strojnog učenja u svrhu pronalaženja novih strategija za pristup navedenoj problematici. Prikazat će se razvoj računalnog modela strojnog učenja za sortiranje otpada tako što će se trenirati već postojeća mrežna arhitektura u programskom jeziku Python uz korištenje potrebnih biblioteka. Prije samog procesa bilo je potrebno proučiti i odabrati adekvatne metode dubokog učenja prikladne za rješavanje zadanog problema. Isto tako, kako su temelj strojnog učenja odgovarajući skupovi podataka, bilo ih je potrebno pronaći, obraditi i organizirati. U ovom slučaju skup podataka predstavlja skup fotografija raznih vrsta otpada. Nakon svega, prikazat će se evaluacija rada modela kroz eksperimente u sklopu Laboratorija za projektiranje izradbenih i montažnih sustava.

Abstract (english)

The global problem of waste management requires creative approaches to simplify sorting procedures. Modern waste streams present unique challenges to conventional approaches, which is why the use of state-of-the-art technologies is necessary. The aim of this paper is to show the possibilities of using the potential of machine learning in order to find new strategies for approaching the mentioned problem. The development of a machine learning computational model for waste sorting will be presented by training an already existing neural network in the Python programming language using the necessary libraries. Before the process itself, it was necessary to study and select adequate deep learning methods suitable for solving the given problem. Also, since the basis of machine learning is the appropriate datasets, it was necessary to find, process and organize them. In this case, the dataset represents a set of photographs of various types of waste. After all, the evaluation of the model's work through experiments as part of the Laboratory for Designing Manufactured and Assembled Systems will be presented.

• Keywords: strojno učenje; duboko učenje; mrežna arhitektura; obuka modela; skupovi podataka; evaluacija; klasifikacija; umjetna inteligencija
• Keywords (english): machine learning; deep learning; network architecture; model training; datasets; evaluation; classification; artificial intelligence
• Language: croatian
• URN:NBN: urn:nbn:hr:235:528266
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka strojarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2023-09-26 23:14:27