On-Device Event Filtering with Binary Neural Networks for Pedestrian Detection Using Neuromorphic Vision Sensors

Keyphrases

• On-device 100%
• Hardware Efficiency 100%
• Event Filtering 100%
• Pedestrian Detection 100%
• Binary Neural Network 100%
• Neuromorphic Vision Sensor 100%
• Latency 50%
• Neural Network 50%
• Light Intensity 50%
• Denoising 50%
• Main Components 50%
• Classification Accuracy 50%
• Dynamic Range 50%
• Low Power Consumption 50%
• Frame-based 50%
• Field Programmable Gate Array Implementation 50%
• Low Bandwidth 50%
• Discrete Changes 50%
• Imaging Sensor 50%
• Image Stream 50%
• Asynchronous Cameras 50%
• High Sampling Frequency 50%
• Temporal Windowing 50%
• Floating-point 50%
• Camera Sensor 50%
• Point-process Filter 50%
• Dynamic Vision Sensor 50%
• Low-complexity Neural Networks 50%
• Bandwidth Limitation 50%
• Neuromorphic Dynamics 50%
• Efficient Architecture 50%

Engineering

• Dynamic Range 100%
• Classification Accuracy 100%
• Field Programmable Gate Arrays 100%
• Input Image 100%
• Main Component 100%
• Floating Point 100%
• Point Process 100%
• Sampling Frequency 100%
• Low Power Consumption 100%
• Bandwidth Requirement 100%
• Luminous Intensity 100%

Computer Science

• Neural Network 100%
• Computer Hardware 50%
• Classification Accuracy 25%
• Floating Point 25%
• Main Component 25%
• Imaging Sensor 25%
• Sampling Frequency 25%
• Low Power Consumption 25%
• Windowing 25%
• Bandwidth Requirement 25%
• Field Programmable Gate Arrays 25%
• Luminous Intensity 25%