MadEdge: Home

MadEdge is an interdisciplinary research group at the University of Wisconsin-Madison that aims to build practical, scalable, and deployable computational (AI-empowered) edge applications. We build holistic hardware-software solutions at in-suit edge sensors, near-edge gateways, and far-edge data centers, which make microsecond, millisecond, and second-level application-defined computation possible. We are eager to collaborate with researchers from different fields and bring cutting-edge hardware infrastructure and software stacks to your problem domains.
MadEdge is developed and maintained by NetLab at the Computer Sciences Department.

Usecases

Global Vegetation Photosynthesis

This project aims to use large-scale remote sensing data to monitor variations in vegetation photosynthetic carbon uptake from hourly to yearly timescales.

Parameterized Earth System Model

This project aims to enhance wildfire simulations in current ESMs and support more effective wildfire management.

Aerial Image Analysis

This project aims to identify individual trees across the United States using high-resolution aerial imagery and deep learning.

Real-time Cow Detection and Tracking

This project develops a distributed video analytics framework for cattle farms. We aim to detect and track cow activities from multiple video streams (30FPS) at millisecond granularity.

Object Detection for HoloLens

This project develops a millisecond-scale object recognition system for Microsoft HoloLens. We aim to increase the service time of the glass device by offloading computation tasks to edge servers without hurting user experience.