An Overview of the SVIRO Dataset and Benchmark

Steve Dias Da Cruz; Oliver Wasenmüller; Hans-Peter Beise; Thomas Stifter; Didier Stricker

In: Computer Science in Cars Symposium. ACM Computer Science in Cars Symposium (CSCS-2019), October 8, Kaiserslautern, Germany, ACM, 2019.


in this extended abstract, we provide an overview of SVIRO, arecently generated synthetic dataset for sceneries in the passengercompartment of ten different vehicles. We showed that SVIRO canbe used to analyze machine learning-based approaches for theirgeneralization capacities and reliability across several tasks whentrained on a limited number of variations (e.g. identical backgroundsand textures, few instances per class). This is in contrast to theintrinsically high variability of common benchmark datasets andas a result SVIRO allows investigations under novel circumstances.


sviro_cscs.pdf (pdf, 2 MB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence