Technology Category
Machine Learning: Learning on the Edge & Model CompressionComputer Vision: Learning & Optimization for CVCognitive Modeling & Cognitive Systems: Neural Spike Coding
EA: An Event Autoencoder for High-Speed Vision Sensing
🤖 AI Summary
To address the sparsity, high noise levels, and computational redundancy inherent in event camera data, this paper proposes an efficient event autoencoder architecture integrating convolutional encoding, adaptive threshold selection, and a lightweight classifier—enabling simultaneous event stream compression/reconstruction and rapid object recognition. The method preserves critical spatiotemporal features while drastically reducing model parameters and computational overhead, facilitating real-time deployment on edge devices. On the SEFD dataset, it achieves detection accuracy comparable to YOLO-v4, with a 35.5× reduction in parameter count. It attains up to 44.8 FPS on both Raspberry Pi and Jetson Nano, delivering an 87.84× speedup over state-of-the-art methods. The core contributions lie in (i) compact representation learning tailored to asynchronous event streams and (ii) an end-to-end lightweight design explicitly optimized for edge computing constraints.
Technology Category
Application Category
📝 Abstract
High-speed vision sensing is essential for real-time perception in applications such as robotics, autonomous vehicles, and industrial automation. Traditional frame-based vision systems suffer from motion blur, high latency, and redundant data processing, limiting their performance in dynamic environments. Event cameras, which capture asynchronous brightness changes at the pixel level, offer a promising alternative but pose challenges in object detection due to sparse and noisy event streams. To address this, we propose an event autoencoder architecture that efficiently compresses and reconstructs event data while preserving critical spatial and temporal features. The proposed model employs convolutional encoding and incorporates adaptive threshold selection and a lightweight classifier to enhance recognition accuracy while reducing computational complexity. Experimental results on the existing Smart Event Face Dataset (SEFD) demonstrate that our approach achieves comparable accuracy to the YOLO-v4 model while utilizing up to $35.5 imes$ fewer parameters. Implementations on embedded platforms, including Raspberry Pi 4B and NVIDIA Jetson Nano, show high frame rates ranging from 8 FPS up to 44.8 FPS. The proposed classifier exhibits up to 87.84x better FPS than the state-of-the-art and significantly improves event-based vision performance, making it ideal for low-power, high-speed applications in real-time edge computing.
Problem
Research questions and friction points this paper is trying to address.
Addresses motion blur and latency in high-speed vision sensing
Improves object detection from sparse, noisy event camera data
Reduces computational complexity while maintaining recognition accuracy
Innovation
Methods, ideas, or system contributions that make the work stand out.
Event autoencoder compresses and reconstructs event data
Convolutional encoding with adaptive threshold selection
Lightweight classifier reduces computational complexity significantly
🔎 Similar Papers
No similar papers found.
