hh.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Explosive detection method for mining industry by two-stage video object detection
Halmstad University, School of Business, Innovation and Sustainability.
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

The mining industry is a perilous environment for its workers. Inmany such locations, dynamite is employed for mine blasting to expedite operations. For instance, if a robot were equipped with a camera instead of a human, and this robot could autonomously identify unexploded dynamite, it could significantly enhance the safetyof mine workers following an explosion. This system would necessitate object detection in conjunction with the robot’s visual perception. This report seeks to address the question: ”Can a ConvolutionalNeural Network (CNN), such as YOLOv8, still accurately predictreal objects when trained solely with semi-synthetic images?” Themethod for this work included detection using a color range on anedge device followed by (to double check) detection using AI withYOLOv8, which was trained with only real background images with3d CAD images of a dynamite using the program Blender pasted ontop. Stage 2 (YOLOv8) after being trained with only semi-syntheticimages could predict with approximately 99.1% accuracy in imagesof plastic dynamite as having a dynamite. Stage 2, despite trainedwith much less epochs than recommended (10 instead of 100) was accurate when being tested on real images with a similar environment(beach) as the semi-synthetic images but not so accurate when testedon real images of a dynamite and a similar object in environmentsnot similar to the training images, in those images there where stillmany detections but mostly with many smaller bounding boxes onthe same dynamite instead of one big bounding box.

Place, publisher, year, edition, pages
2024.
National Category
Computer Systems
Identifiers
URN: urn:nbn:se:hh:diva-54271OAI: oai:DiVA.org:hh-54271DiVA, id: diva2:1883341
Educational program
Master's Programme in Embedded and Intelligent Systems, 120 credits
Available from: 2024-08-08 Created: 2024-07-09 Last updated: 2025-10-01Bibliographically approved

Open Access in DiVA

fulltext(17925 kB)714 downloads
File information
File name FULLTEXT02.pdfFile size 17925 kBChecksum SHA-512
c6393a57874786cac41c5ebd2902737f6cd3e3abb8ada11504b7030fd59e5f3c71b442c54cada88a27e531d607d0abf4d6219e2b4debad63e5e64c80f15aedfd
Type fulltextMimetype application/pdf

By organisation
School of Business, Innovation and Sustainability
Computer Systems

Search outside of DiVA

GoogleGoogle Scholar
Total: 715 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 511 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf