Page - 121 - in Joint Austrian Computer Vision and Robotics Workshop 2020

a previous implementation of the CarCutter service. Direct comparisonofmasksgeneratedbyour imple- mentationwithonesgeneratedbytheserviceinApril of2019 showeda reductionof65% in thesegmenta- tionerrorasmeasuredbytheJaccardindex. With the exception of our postprocessing procedure the pre- sented methods are applicable to the general image segmentation task. Acknowledgments We would like to thank micardo GmbH3 for a fruitful collaboration and the use of their private dataset. This work has been partly funded by the Austriansecurity researchprogrammeKIRASof the FederalMinistryforTransport, InnovationandTech- nology (bmvit)underGrant873495. References [1] V.Badrinarayanan,A.Kendall, andR.Cipolla. Seg- net: Adeepconvolutionalencoder-decoderarchitec- ture for image segmentation. 2017 IEEE Transac- tions on Pattern Analysis and Machine Intelligence (TPAMI), 2017. [2] M.Cordts,M.Omran,S.Ramos,T.Rehfeld,M.En- zweiler, R. Benenson, U. Franke, S. Roth, and B. Schiele. The cityscapes dataset for semantic ur- ban scene understanding. In 2016 IEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages3213–3223, June2016. [3] D. Eigen and R. Fergus. Predicting depth, surface normals and semantic labels with a common multi- scale convolutional architecture. In Proceedings of the 2015 IEEE International Conference on Com- puter Vision (ICCV), ICCV ’15, pages 2650–2658, Washington,DC,USA,2015. IEEEComputerSoci- ety. [4] Feng Ning, D. Delhomme, Y. LeCun, F. Piano, L. Bottou, and P. E. Barbano. Toward automatic phenotyping of developing embryos from videos. 2005IEEETransactionsonImageProcessing(TIP), 14(9):1360–1371,Sep.2005. [5] K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. 2016 IEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages770–778,2016. [6] V. Iglovikov and A. Shvets. Ternausnet: U-net with VGG11 encoder pre-trained on imagenet for image segmentation. CoRR, abs/1801.05746, 2018. [7] M. Kass, A. Witkin, and D. Terzopoulos. Snakes: Active contour models. International Journal of ComputerVision, 1(4):321–331,1988. 3https://www.micardo.com/ (accessedFebruary24, 2020) [8] M. Khened, V. Alex Kollerathu, and G. Krish- namurthi. Fully convolutional multi-scale resid- ual densenets for cardiac segmentation and auto- mated cardiac diagnosis using ensemble of classi- fiers. Medical Image Analysis, 51, 012018. [9] F. Milletari, N. Navab, and S. Ahmadi. V-net: Fully convolutional neural networks for volumetric med- ical image segmentation. In 2016 Fourth Interna- tional Conference on 3D Vision (3DV), pages 565– 571,Oct2016. [10] C. Rhemann, C. Rother, J. Wang, M. Gelautz, P. Kohli, and P. Rott. A perceptually motivated on- line benchmark for image matting. In 2009 IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), pages 1826–1833, June2009. [11] O. Ronneberger, P.Fischer, and T. Brox. U-net: Convolutional networks for biomedical image seg- mentation. In 2015 Medical Image Computing andComputer-AssistedIntervention(MICCAI), vol- ume9351ofLNCS, pages234–241.Springer,2015. (availableon arXiv:1505.04597 [cs.CV]). [12] P. Sermanet, D. Eigen, X. Zhang, M. Mathieu, R. Fergus, and Y. Lecun. Overfeat: Integrated recognition, localization and detection using convo- lutionalnetworks. In2014InternationalConference on Learning Representations (ICLR), CBLS, April 2014, 2014. [13] E. Shelhamer, J. Long, and T. Darrell. Fully convo- lutional networks for semantic segmentation. 2017 IEEETransactionsonPatternAnalysisandMachine Intelligence (TPAMI), 39(4):640–651,Apr.2017. [14] K. Simonyan and A. Zisserman. Very deep convo- lutional networks for large-scale image recognition. In 2015 International Conference on Learning Rep- resentations (ICLR), 2015. [15] J. Xu, H. Guo, A. Kageza, S. Wu, and S. AlQarni. Removing background with semantic segmentation basedonensemble learning. EAI, 92018. [16] N. Xu, B. Price, S. Cohen, and T. Huang. Deep im- agematting. In2017IEEEConferenceonComputer VisionandPatternRecognition(CVPR), pages311– 320, July 2017. 121