Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Resolving Classifcation Ambiguities in Convolutional Neural Networks Using Hierarchical Structures

: Albadawi, Mohamad
: Lukas, Uwe von; Krause, Tom

Volltext urn:nbn:de:0011-n-5126055 (14 MByte PDF)
MD5 Fingerprint: 84af733480638dd7456516ac857e20ad
Erstellt am: 9.10.2018

Rostock, 2018, 72 S.
Rostock, Univ., Master Thesis, 2018
Master Thesis, Elektronische Publikation
Fraunhofer IGD ()
Visual Computing as a Service; Computer vision (CV); Object detection; Convolutional Neural Networks (CNN); Deep learning; improved object classification

We have recently witnessed the revolution of deep learning and convolutional neural networks enabled by the powerful machines available today. Convolutional neural networks have demonstrated excellent performance on various vision tasks, most importantly classification and detection. Nevertheless, there are some difficulties in the way of perfect performance. One problem is discriminating among objects that look extremely similar visually but semantically they are different. Another problem is the high cost of training large detection models. The same cost applies when the model is required to detect a new type of objects. In this work those problems are handled by introducing visual concepts and the use of hierarchical structures. We will see how the accuracy of classifying similar objects can be highly improved and how the time of accommodating for new objects in a detection model can be reduced from days to hours.