ParslAI - Actionable Aerial Analytics

Why?

More predictable than weather

adverse atmospheric conditions are a given and are outside of human control

Bad weather in - bad insights out

machine learning workflow optimization is the aerial imaging value frontier

Vast amount of geospatial and aerial imagery has been collected and continues to be generated daily. The image acquisition, storage and transmission constraints, particularly for mission-critical applications, require robust end-to-end optimization. Our ability to effectively extract deep, actionable planetary scale insights is key to driving real outcomes in addressing sustainable development goals.

100s of terabytes
imaged daily

Repeat fly-overs

Storage and
transmission

Our Solution

Best-in-category, Generalizable, Optimized end-to-end computer vision

Heavy Haze Challenge

Color and Visibility Restoration

Target Detection Optimization

Publications

S. A. Hovhannisyan, H. A. Gasparyan, S. S. Agaian and A. Ghazaryan,
"AED-Net: A single image dehazing,"
in IEEE Access, doi: 10.1109/ACCESS.2022.3144402.
(Published in Jan. 2022)

READ PAPER

H. A. Gasparyan, S. A. Hovhannisyan, S. V. Babayan, and S. S. Agaian,
"Iterative Retinex-Based Decomposition Framework for Low Light Visibility Restoration,"
in IEEE Access, doi: 10.1109/ACCESS.2023.3269719.
(Published in April 2023)

READ PAPER