@inproceedings{FichtelFruehwaldHoeschetal., author = {Fichtel, Lars and Fr{\"u}hwald, Alexander and H{\"o}sch, Leonhard and Schreibmann, Vitaliy and Bachmeir, Christian}, title = {Tree Localization and Monitoring on Autonomous Drones employing Deep Learning}, series = {PROCEEDING OF THE 29TH CONFERENCE OF FRUCT ASSOCIATION}, booktitle = {PROCEEDING OF THE 29TH CONFERENCE OF FRUCT ASSOCIATION}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:863-opus-24908}, abstract = {Forest management relies on the analysis of satellite imagery and time intensive physical on-site inspections. Both methods are costly and time consuming. Satellite based images are often not updated in a sufficient frequency to react to infestations or other occurring problems. Forest management benefits greatly from accurate and recent information about the local forest areas. In order to react appropriately and in time to incidents such as areas damaged by storms, areas infested by bark beetles and decaying ground water level, this information can be extracted from high resolution imagery. In this work, we propose UAVs to meet this demand and demonstrate that they are fully capable of gathering this information in a cost efficient way. Our work focuses on the cartography of trees to optimize forest-operation. We apply deep learning for image processing as a method to identify and isolate individual trees for GPS tagging and add some additional information such as height and diameter.}, language = {en} } @inproceedings{FichtelErbacherHelleretal., author = {Fichtel, Lars and Erbacher, Dominik and Heller, Leon and Fr{\"u}hwald, Alexander and H{\"o}sch, Leonhard and Bachmeir, Christian}, title = {Analysis of Object Detection Datasets for Machine Learning with Small and Tiny Objects}, series = {Proceedings of the Eleventh International Conference on Engineering Computational Technology}, booktitle = {Proceedings of the Eleventh International Conference on Engineering Computational Technology}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:863-opus-24895}, abstract = {Deep Learning models are trained to detect humans, cars, and other large objects which are centered in the images. The same models struggle with detecting small and tiny objects because of architecture design decisions that reduce the entropy of small and tiny objects during training. These small and tiny objects are essential for damage identification and maintenance including inspection and documentation of aeroplanes, constructions, offshore structures, and forests. Our work defines the terms tiny and small in context of deep learning models to evaluate possible approaches to resolve the issue of low accuracy in detecting these objects. We analyse the currently applied common datasets Common Objects in Context, ImageNet and Tiny Object Detection Challenge dataset. In addition we compare these datasets and present the differences in terms of object instance size. The COCO dataset, ImageNet dataset and TinyObjects dataset are analysed regarding size categorization and relative object size. The results show the large differences between the size ratios of the three chosen datasets, with ImageNet having by far the largest object instances, COCO being in the middle and TinyObjects having the smallest objects as its name would indicate. Since the objects themselves are larger in terms of total pixel width and height, they therefore make up a bigger percentage on the superordinate picture. Looking at the size categories of the COCO dataset and our extension of the tiny and very small category, the results confirm the size hierarchy of the datasets. With ImageNet having most of its objects in the large category, COCO respectively in the medium category and TinyObjects in the very small category. By taking these results into account, the reader is able to choose a fitting dataset for their tasks.We expect our analysis to help and improve future research in the area of small and tiny object detection.}, language = {en} } @incollection{FichtelErbacherGruenwaldetal., author = {Fichtel, Lars and Erbacher, Dominik and Gr{\"u}nwald, Dennis and Heller, Leon and Bachmeir, Christian and Timofte, Radu}, title = {TASOD: A Data Collection for Tiny and Small Object Detection}, series = {Lecture Notes in Computer Science}, booktitle = {Lecture Notes in Computer Science}, publisher = {Springer Nature Switzerland}, address = {Cham}, isbn = {9783031918551}, issn = {0302-9743}, doi = {https://doi.org/10.1007/978-3-031-91856-8_14}, pages = {229 -- 245}, language = {en} } @article{Bachmeir, author = {Bachmeir, Christian}, title = {Mikrologistik der Zukunft mit dezentral organisierten boden- und luftgebundenen autonomen F{\"o}rdereinheiten}, series = {FHWS Science Journal}, volume = {5}, journal = {FHWS Science Journal}, number = {2}, issn = {2196-6095}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:863-opus-20086}, pages = {117 -- 124}, abstract = {Gepr{\"a}gt durch den Einzug der Digitalisierung in Wirtschaft und Industrie sieht sich die Logistikbranche heute zunehmend damit konfrontiert, immer kleinere St{\"u}ckmengen und Losgr{\"o}ßen transportieren zu m{\"u}ssen. Heutige, konventionelle Logistiksysteme sind f{\"u}r diese zuk{\"u}nftigen Anforderungen nicht ausgelegt. Deshalb bringt der Transport kleiner Losgr{\"o}ßen hohe Kosten mit sich. Dar{\"u}ber hinaus sind heutige Logistiksysteme nicht flexibel genug. Auch die praktizierte Ansammlung von kleinen Losgr{\"o}ßen zu gr{\"o}ßeren Einheiten erf{\"u}llt die Anforderungen nicht. Der Warentransport wird aus Sicht des steigenden Anspruchs an den Zeitfaktor teuer. Dar{\"u}ber hinaus entstehen zus{\"a}tzliche Lagerkosten. Neue Technologien wie fahrerlose Transportsysteme und Transportdrohnen k{\"o}nnen in ihrer aktuellen Entwicklung dem Bedarf der Mikromobilit{\"a}t nur in stark abgegrenzten Bereichen gerecht werden und somit ihr Potenzial noch nicht ganzheitlich entfalten. Im Projekt wird die kollaborative Zusammenarbeit verschiedener Transporteinheiten wie fahrerlose Transportsysteme (FTS), Multicopter- und Fl{\"a}chendrohnen aufbauend auf einem KI-Service entwickelt, was einem heterogenen Netzwerk die {\"U}berwindung der einzelnen, technologischen Systemgrenzen und zweckdienlichen Einsatz erm{\"o}glicht. Die Auslegung von logistischen Systemen erfolgt immer nach dem Prinzip des minimalsten Aufwandes. Das Projekt FlowPro entwickelt ein auf KI-Verfahren basierendes Logistiknetzwerk, welches sich selbstst{\"a}ndig organisiert und in Industrieparks die Intralogistik sowie dar{\"u}ber hinaus unternehmens{\"u}bergreifend auf dem Land- und Luftweg die Mikromobilit{\"a}t von Waren erm{\"o}glicht und optimiert.}, language = {de} } @article{Bachmeir, author = {Bachmeir, Christian}, title = {Das Projekt KI-Inspektionsdrohne}, series = {FHWS Science Journal}, volume = {5}, journal = {FHWS Science Journal}, number = {2}, issn = {2196-6095}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:863-opus-19994}, pages = {131 -- 134}, abstract = {Das Verbundvorhaben KI-Inspektionsdrohne, im Rahmen des Luftfahrtforschungsprogramms V [1], hat als Ziel Maintenance-, Repair- und Overhaul-(MRO-)Prozesse in der Luftfahrtindustrie zu beschleunigen, und damit die MRO-Industrie noch wettbewerbsf{\"a}higer zu machen. Im Vorhaben wird ein sicheres System zur Schadensdetektion und -bewertung von {\"a}ußeren Flugzeugstrukturen unter Ber{\"u}cksichtigung aktueller Instandhaltungs-Anforderungen der Luftfahrtbranche entwickelt. Der entwickelte Prototyp integriert vernetzte UAV (Unmanned Aerial Vehicles - Unbemannte Flugsysteme), autonome Navigation und Schadensaufnahme und KI-gest{\"u}tzte Auswertung und stellt ein Decision Support System zur Verf{\"u}gung bzw. f{\"a}llt eigenst{\"a}ndig Entscheidungen. Forschungsschwerpunkt der FHWS ist, bzw. Forschungsschwerpunkte sind industriespezifische Ende-zu-Ende-Security, KI, Safety, und insbesondere die Absicherung der mobilen, mit Cloud oder Edge verbundenen Einheiten gegen Cyberattacken.}, language = {de} }