Hinweis Dies ist nicht die aktuellste Version der Ressource. Diese kann hier gefunden werden: https://fordatis.fraunhofer.de/handle/fordatis/463.2
Langanzeige der Metadaten
DC ElementWertSprache
dc.contributor.advisorEmmert, Johannes-
dc.contributor.advisorLeisner, Johannes-
dc.contributor.advisorMaier, Georg-
dc.contributor.authorRoming, Lukas-
dc.contributor.authorRüger, Steffen-
dc.contributor.authorKalb, Michael-
dc.contributor.authorBeutler, Frederik-
dc.contributor.otherTavoosi, Kimia-
dc.contributor.otherJosekutty, Jerardh-
dc.contributor.otherGhosh, Tridib-
dc.contributor.otherLauerer, Franziska-
dc.contributor.otherTrihasworo, Rio Hilmawan Bagas-
dc.contributor.otherBhatti, Zenib-
dc.contributor.otherWein, Floria Simone-
dc.contributor.otherShubhangi, Shubhangi-
dc.date.accessioned2026-01-07T10:19:25Z-
dc.date.available2026-01-07T10:19:25Z-
dc.date.issued2025-12-
dc.identifier.urihttps://fordatis.fraunhofer.de/handle/fordatis/463-
dc.identifier.urihttp://dx.doi.org/10.24406/fordatis/420-
dc.description.abstractThe dataset comprises 586 RGB images of post-consumer lightweight packaging waste captured on a conveyor belt. Annotations follow the COCO format. The samples were kindly provided by a recycling plant of Lobbe RSW GmbH in Iserlohn, Germany. Samples were recorded as received, with adherent contaminants and without pretreatment. This dataset is part of the multi-sensor-dataset of the project K3I-Cycling and funded by the Federal Ministry of Research, Technology and Space (BMFTR), Germany under the funding reference 033KI201.en
dc.description.sponsorshipThis dataset is part of the multi-sensor-dataset of the project K3I-Cycling and funded by the Federal Ministry of Research, Technology and Space (BMFTR), Germany under the funding reference 033KI201.en
dc.language.isoenen
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/en
dc.subjectPackaging Wasteen
dc.subjectBenchmark Dataseten
dc.subjectImagingen
dc.subjectInstance Segmentationen
dc.subjectObject Detectionen
dc.subjectAutomated Waste Sortingen
dc.subjectSensor-based Sortingen
dc.subjectOptical Sortingen
dc.subject.ddcDDC::600 Technik, Medizin, angewandte Wissenschaften::600 Technik::600 Technik, Technologieen
dc.titlePackWISE: Packaging Waste Instance Segmentation Dataset for Model Training and Evaluationen
dc.title.alternativeRGB Images of Post Consumer LWP Waste in a realistic Sorting Plant Settingen
dc.typeImageen
dc.contributor.funderBundesministerium für Bildung und Forschung BMBF (Deutschland)en
dc.description.technicalinformationImages are provided as JPEG files of size 4096x4096 pixel. Annotations follow the COCO format: each instance is delineated by a pixel-level mask, a bounding-box, and assigned to one of 23 categories. Bounding box coordinates are given as Integers. Pixel masks are binary run-length encoded (RLE) and can be decoded using pycocotools. Annotations can be converted to YOLO format (bounding boxes only) using the python script convert_dataset_to_yolo_format.py in the root folder of the dataset. The dataset is divided into three splits: training (70%), validation (15%), and test (15%). Image acquisition employed a prism-based RGB line-scan camera (SW-4000T-10GE) with a resolution of 4096 pixels. The line rate was set to 1278 lines per second. Two-dimensional images were reconstructed by scanning the samples as they moved along a conveyor belt at 0.2 m/s. Illumination was provided by 12 halogen lamps.en
dc.title.translatedPackWISE: Datensatz zur Instanzsegmentierung von Verpackungsabfällen für Training und Evaluation von Modellende
fordatis.groupMikroelektroniken
fordatis.instituteIOSB Fraunhofer-Institut für Optronik, Systemtechnik und Bildauswertungen
fordatis.instituteIIS Fraunhofer-Institut für Integrierte Schaltungenen
fordatis.project.fhgid11-04441-2140-00001en
fordatis.project.fhgid11-04441-2440-00018en
fordatis.rawdatafalseen
fordatis.sponsorship.FundingProgrammeKI-Anwendungshub Kunststoffverpackungen – nachhaltige Kreislaufwirtschaft durch Künstliche Intelligenzen
fordatis.sponsorship.projectid033KI201en
fordatis.sponsorship.projectnameKI-gestützte Optimierung der Kreislaufführung von Kunststoffverpackungenen
fordatis.sponsorship.projectacronymK3I-Cyclingen
fordatis.sponsorship.ResearchFrameworkProgrammKI-Strategie der Bundesregierungen
fordatis.sponsorship.ResearchFrameworkProgrammForschung für Nachhaltigkeit (FONA)en
fordatis.date.start2024-07-09-
fordatis.date.end2024-07-12-
Enthalten in den Sammlungen:Fraunhofer-Institut für Optronik, Systemtechnik und Bildauswertung IOSB

Dateien zu dieser Ressource:
Datei Beschreibung GrößeFormat 
PackWISE_dataset_v1.zip354,5 MBZIPÖffnen/Download
Zur Kurzanzeige

Versionshistorie
Version Ressource Datum Zusammenfassung
2 fordatis/463.2 2026-02-25 09:22:22.867
1 fordatis/463 2026-01-07 11:19:25.0
Ausgewählte Version

Diese Ressource wurde unter folgender Copyright-Bestimmung veröffentlicht: Lizenz von Creative Commons Creative Commons