Detekcija paketa na policama u uvjetima smanjene vidljivosti

undergraduate thesis

• Title (english): Detection of packages on shelves in conditions of reduced visibility
• Author: Lana Barišić
• Mentor: Tamara Petrović (mentor)
• Committee member: Tamara Petrović (član povjerenstva)
• Committee member: Marko Križmančić (član povjerenstva)
• Committee member: Robert Milijaš (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad bavi se razvojem algoritma za detekciju paketa s QR kodovima na policama u uvjetima smanjene vidljivosti koristeći bespilotne letjelice s kamerama te s mogućnošću dodatnog osvjetljenja. Algoritam je testiran u različitim razinama ambijentalne svjetlosti. Postignuti rezultati pokazuju visoku točnost modela detekcije kartonskih kutija, s mPA vrijednošću od 0.968 nakon 157 epoha u prvom treningu. Model je pokazao dosljednost predikcija i u stvarnim situacijama, ali nailazi na poteškoće prilikom dekodiranja QR kodova. U radu su objašnjeni osnovni pojmovi potrebni za shvaćanje izrade modela, nakon čega su prezentirani dobiveni rezultati.

Abstract (english)

This work deals with the development of an algorithm for package detection with QR codes on shelves under low visibility conditions using drones equipped with cameras with additional lighting capabilities. The algorithm was tested under different levels of ambient light. The achieved results demonstrate a high accuracy of the model, with an mPA value of 0.968 after 157 epochs in the first training session. The model showed consistency in predictions even in real-world situations, but encounters difficulties when decoding QR codes. The work explains the basic concepts necessary for understanding the model development, followed by the presentation of the obtained results.

• Keywords: računalni vid; konvolucijske neuronske mreže; detekcija objekata
• Keywords (english): computer vision; convolutional neural networks; object detection
• Language: croatian
• URN:NBN: urn:nbn:hr:168:983996
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2024-10-12 20:04:02