Object detection in factory based on deep learning approach

Manufacturing systems need to be highly adaptive and flexible to meet heterogeneous customer requirements and require fast and accurate object identification. This paper presents the implementation of deep learning for the object detection in a factory in which images of machines are used to train f...

Full description

Saved in:
Bibliographic Details
Published inProcedia CIRP Vol. 104; pp. 1029 - 1034
Main Authors Yi, Li, Siedler, Carina, Kinkel, Yann, Glatt, Moritz, Kölsch, Patrick, Aurich, Jan C.
Format Journal Article
LanguageEnglish
Published Elsevier B.V 2021
Subjects
convolutional neural network
deep learning
manufacturing systems
object detection
supervised learning
deep learning
manufacturing systems
object detection
supervised learning
convolutional neural network
Online AccessGet full text

Cover

More Information
Summary:Manufacturing systems need to be highly adaptive and flexible to meet heterogeneous customer requirements and require fast and accurate object identification. This paper presents the implementation of deep learning for the object detection in a factory in which images of machines are used to train four models. For each model, three trials with different hyperparameter configurations are performed. After the training, evaluations based on the loss curves, mean average precision curves, and performance metrics are carried out. Finally, the trained models are tested with a video in order to select the model with the best detection performance and accuracy.
ISSN:2212-8271
2212-8271
DOI:10.1016/j.procir.2021.11.173