English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Due to the high activation sparsity and use of accumulates (AC) instead ofexpensive multiply-and-accumulates (MAC), neuromorphic spiking neural networks(SNNs) have emerged as a promising low-power alternative to traditional DNNsfor several computer vision (CV) applications. However, most existing SNNsrequire multiple time steps for acceptable inference accuracy, hinderingreal-time deployment and increasing spiking activity and, consequently, energyconsumption. Recent works proposed direct encoding that directly feeds theanalog pixel values in the first layer of the SNN in order to significantlyreduce the number of time steps. Although the overhead for the first layer MACswith direct encoding is negligible for deep SNNs and the CV processing isefficient using SNNs, the data transfer between the image sensors and thedownstream processing costs significant bandwidth and may dominate the totalenergy. To mitigate this concern, we propose an in-sensor computinghardware-software co-design framework for SNNs targeting image recognitiontasks. Our approach reduces the bandwidth between sensing and processing by12-96x and the resulting total energy by 2.32x compared to traditional CVprocessing, with a 3.8% reduction in accuracy on ImageNet.

In-Sensor & Neuromorphic Computing are all you need for Energy Efficient Computer Vision