Automated Image Anomaly Detection System

AIADS uses multimodal geospatial data, identifies and classifies unusual changes, and monitors vast areas on a daily basis.

Manual analysis doesn’t scale

Existing solutions rely on manual analysis, which is slow, expensive and does not scale. Customers need automation to analyze images for unusual activity and prioritize the areas of interest.

AIADS automates image analysis and prioritizes images with unusual change.

Illustration of the globe with icons for servers, users and analysts

How it fits your workflow

Input

Satellites and drones: optical, multispectral and SAR imagery.

AIADS

AI compares each new image with the historical record of the same area and separates normal change from unusual change.

Output

Anomaly files and notifications.

AIADS features

Turnkey solution

Ready to run on imagery you already collect.

Fits your workflow

Outputs load into the tools your analysts use.

Fully automated

No manual image-by-image review to find change.

Resolution agnostic

Works from high-resolution commercial imagery to free Sentinel data.

Replace change detection plus manual analysis

AIADS detects anomalies, not just change. It’s faster, smarter and scalable.

Existing options

Change detection is not completely automated
Noisy, with too many false positives
Manual analysis is time consuming
Results are delayed
Neither approach scales

Simularity temporal change detection

Needs minimal analyst input
Flags only anomalies
Notifies immediately
Easily scalable, with low capital cost

Stack of satellite images of one Houston site over time, with an anomaly heat map on the newest

Temporal change detection uses historical context

Detecting construction and land clearing in Houston, Texas. The candidate image from August 22, 2017 was analyzed against earlier images from May 3, April 8 and January 12, 2017 and November 15, 2016. The heat map overlay marks the unusual changes.