Abstract
Neuromorphic Vision sensors have improved greatly since the first silicon retina was presented almost three decades ago. They have recently matured to the point where they are commercially available and can be operated by laymen. However, despite improved availability of sensors, there remains a lack of good datasets, while algorithms for processing spike-based visual data are still in their infancy. On the other hand, frame-based computer vision algorithms are far more mature, thanks in part to widely accepted datasets which allow direct comparison between algorithms and encourage competition. We are presented with a unique opportunity to shape the development of Neuromorphic Vision benchmarks and challenges by leveraging what has been learnt from the use of datasets in frame-based computer vision. Taking advantage of this opportunity, in this paper we review the role that benchmarks and challenges have played in the advancement of frame-based computer vision, and suggest guidelines for the creation of Neuromorphic Vision benchmarks and challenges. We also discuss the unique challenges faced when benchmarking Neuromorphic Vision algorithms, particularly when attempting to provide direct comparison with frame-based computer vision.
Highlights
Benchmarking using widely accepted datasets is important for algorithm development
Computer Vision (CV) is an obvious example where open access to good datasets has been integral in rapid development and maturation of the field (Kotsiantis et al, 2006)
We use the term “Computer Vision” (CV) to denote the conventional approach to visual sensing, which begins with acquisition of images, or sequences of images
Summary
Benchmarking using widely accepted datasets is important for algorithm development. Such benchmarking allows quantitative performance evaluation and comparison between algorithms, promoting competition and providing developers with tangible state-of-the-art targets to beat. Each image is a regular grid of pixels, each pixel having an intensity or color value Such images are a widely accepted, and largely unquestioned first step in visual sensing. Guidelines stated in this paper extend to benchmarks and datasets for other NV sensors once they do reach such a level of maturity. For temporal contrast NV sensors (hereafter referred to as “NV sensors”), each pixel generates an event whenever its change in log-intensity over time exceeds a programmable threshold. In these sensors, any pixel can generate an event at any time, thereby accurately recording when and where scene changes occur. In this article we discuss these challenges (Section 2) and assess the current state of NV datasets (Section 3) before reviewing the role datasets have played in CV (Section 4), identifying valuable lessons (Section 5) and how NV can benefit from these lessons (Section 6)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
