Rutinowski, JeromeJeromeRutinowskiPionezewski, ChristianChristianPionezewskiChilla, TimTimChillaReining, ChristopherChristopherReiningTen Hompel, MichaelMichaelTen Hompel2022-03-152022-03-152021https://publica.fraunhofer.de/handle/publica/41344810.1109/ETFA45728.2021.9613250This paper contributes a new dataset, namely for there-identification of Euro-pallet pallet-blocks, called pallet-block502. Based on a logistics use case, three re-identification algorithms are benchmarked on this dataset. The dataset constitutes502 pallet-blocks, of which ten pictures each are taken, expanding the dataset to a grand total of 5,020 images. The preliminary results of this work indicate the reliable re-identification of pallet-blocks using the Part-based Convolutional Baseline (PCB)network with ResNet50 as its backbone network, achieving an mAP of 98%.entag-freetraceabilityfingerprint of thingsre-identificationcomputer visiondeep learninglogistics658338conference paper