Reproduce “Instantaneous Stereo Depth Estimation of Real-World Stimuli with a Neuromorphic Stereo-Vision Setup” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nicoletta Risi, Enrico Calabrese, Giacomo Indiveri
A neuromorphic stereo-vision system, developed by researchers at the University of Zurich and ETH Zurich, provides real-time, coarse depth estimation from real-world event data. Implemented on mixed-signal neuromorphic hardware, it achieved high accuracy (98-99% correct disparities) with power consumption in the microwatt range.
Key insights
- Mixed-signal neuromorphic hardware can robustly perform real-time, coarse depth estimation from complex, real-world event data despite its inherent analog noise and variability.
- The SNN's coincidence and disparity-sensitive neurons effectively encode stimulus depth instantaneously, adapting their spiking activity to reflect varying depth changes.
- Neuromorphic event-based stereo vision offers substantial energy efficiency (microwatt range) compared to traditional frame-based approaches, making it suitable for energy-constrained applications.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.