Am 2. und 3. Juni erfolgt ein TUdatalib Upgrade auf eine neue Softwareversion. Während dieses Zeitraums steht das System nicht zur Verfügung. Weitere Informationen unter https://www.ulb.tu-darmstadt.de/die_bibliothek/aktuelles/news/news_details_80768.de.jsp. // A TUdatalib upgrade to a new software version is scheduled for June 2nd and 3rd. The system will not be available during that period. Please see https://www.ulb.tu-darmstadt.de/die_bibliothek/aktuelles/news/news_details_80768.en.jsp for further information.
Dataset for automated material flow characterization of shredded WEEE: RGB-camera-based object images and mass data
Description
Three datasets containing data from particles of shredded WEEE, including ferrous metals, non-ferrous metals, plastics and printed circuit boards in two particle size ranges of 12.5 mm - 25 mm and 25 mm - 50 mm, split into image data and mass data.
- Dataset 1 contains images that were used to train and test convolutional neural networks to identify the four material types through image classification, object detection, and instance segmentation. Additionaly, the total mass per material type and particle size range is included.
- Dataset 2 contains images and corresponding particle masses that were used in the training and testing of regression models for particle mass prediction.
- Dataset 3 contains images of particles from three predefined mixed samples, together with the total mass per material type and particle size range in each sample.
The images were recorded with an industry-sized sensor-based sorting machine (Sesotec Varisort Compact [Schoenberg, Germany]) at the pilot-scale sorting plant at Fraunhofer IWKS in Alzenau. Computer vision was used to extract individual particles from the recorded image stream.
The masses were recorded using a precision balance (KERN & SOHN EWJ 3000-2 [Balingen, Germany]) at the output conveyor.
Subject
WEEE;Shredded e-waste;Electronic waste;Machine learning;Material type identification;Particle mass prediction;Sensor-based material flow characterization;Image data;Particle masses;Ferrous metals, non-ferrous metals, plastics, printed circuit boardsDFG subject classification
4.21-03 Mechanische Verfahrenstechnik4.43-04 Künstliche Intelligenz und Maschinelle Lernverfahren
URI
https://tudatalib.ulb.tu-darmstadt.de/handle/tudatalib/4560https://doi.org/10.48328/tudatalib-1743
Collections
The following license files are associated with this item:
Except where otherwise noted, this item's license is described as Creative Commons Attribution 4.0