We present four different robust transfer learning strategies for robust mobile scene recognition . Fine-Tuning in combination withextensive data augmentation improves accuracy and robustness in mobile robot place recognition . We achieved state-of-the-art results using variousbaseline convolutional neural networks and showed the robustness againstlighting and viewpoint changes in challenging mobile robot places recognition . Also, the impact of inferenceoptimization techniques on the general performance is evaluated . We show the generalizationability of our transfer . learning strategies using the KTH-Idol2 database. Also, we tested therobustness of our strategies under viewpoint and lightingchanges .

Author(s) : Hermann Baumgartl, Ricardo Buettner

Links : PDF - Abstract

Code :

Keywords : mobile - strategies - recognition - robust - transfer -

Leave a Reply

Your email address will not be published. Required fields are marked *