Presenter(s)

Wes Baldwin

Download

Download Project (2.6 MB)

Description

This poster presents recent work in the implementation of dimensionality reduction for neuromorphic camera data using time-surfaces. Neuromorphically inspired cameras can operate at extremely high temporal resolution (>800kHz), low latency (20 microseconds), wide dynamic range (>120dB), and low power (30mW). Time-surfaces are an ideal tool to leverage machine learning on event camera datasets as they assist in noise removal while retaining a high degree of spatial and temporal information. Combining time-surfaces with transfer learning is advancing state-of-the-art performance for object classification.

Publication Date

4-24-2019

Project Designation

Independent Research

Primary Advisor

Vijayan K. Asari

Primary Advisor's Department

Electrical and Computer Engineering

Keywords

Stander Symposium project

Object Classification using Neuromorphic Cameras

Share

COinS