DEEPSKY SYSTEM

Terrestrial Meteorological, Atmospheric and Solar Measurements
Power by Vision AI

THE SYSTEM
deepskylogo

Where Atmospheric Physics Meet Image Processing

The DeepSky system focuses on meeting high standards of analysis, quality, and validity. To achieve this, all the theories on Atmospheric Physics are exploited in combination with deep learning and image processing methods in order to obtain the best possible results.

Vision AI System

Novel system based on the technology of commercial cameras, combined with Computer Vision and deep learning techniques for the analysis of digital all-sky images.

Cost Efficiency

Innovation in reducing the cost of terrestrial measurements while maintaining quality and ease of use.

Visible and Infrared Spectrum

Combines visible and infrared images, estimates from propagation models and Vision AI, to provide geophysical variables, both day and night.

GOALS

DeepSky Project Timeline

The main goal of the DeepSky project is to develop an innovative and flexible ground-based measurement system designed to address the needs of end-users in the meteorological, atmospheric and solar energy communities. 

1.

Identify potential end-users, requirements, and market analysis

2.

DeepSky prototype development, calibration, and validation

3.

Algorithmic development for meteorological, atmospheric and solar measurements' processing of different geophysical variables

4.

Development of computer vision and deep learning techniques for cloud type identification, cloud coverage estimation and prediction

5.

System performance evaluation in real-world conditions

DATASET

All-sky Visible Dataset

The dataset comprises a large number of more than 7000 sky images that have been annotated and classified to 7 cloud types. Additional multimodal data (like temperature, humidity, irradiance, particle pollution, etc.) and different periods of time (i.e., different months/years) have been considered in the training, resulting in an overall accuracy of more than 90%.

Cumulus

Puffy clouds with clearly defined edges, white or light gray

Cumulus Sky
Cumulus Sky
Altocumulus

Patched clouds of small cloudlets, mosaic-like white

altocumulus Sky
altocumulus Sky
Cirrus

Thin clouds, fibrous, white feathery clouds of ice crystals or sky covering, whitish

cirrus sky
cirrus sky
Clear Sky

No cloudiness or very few cloudiness, blue

clear sky
Stratocumulus Lumpy layer of clouds, broken to almost overcast, causes fog or fine precipitation, white or gray
Stratocumulus Sky
Stratocumulus Sky

Cumulonimbus

Dark, thick clouds, mostly overcast, indistinct outline, thundercloud or rain cloud, gray
Cumulonimbus sky
Cumulonimbus sky
Mixed

More than two genera of clouds

Mixed Sky
Geometrical calibration of an all-sky imager: from spherical to cartesian and cylindrical coordinates
Time Lapse video capturing the process of external geometric calibration of the sky camera
VISION AI PLATFORM

Complete Vision AI Infrastructure

DeepSky is integrated with PerCV.ai, Irida Labs’ end-to-end software and services platform for deploying Vision AI at scale.

Vision AI sensor

Our Role

Computer vision and deep learning techniques for the analysis of digital all-sky images (cloud types and cloud coverage, cloud prediction etc.)

powered by
Trifold Percv.ai
GET IN TOUCH

Contact Us





    deepskylogo

    Funding Details​

    This project has been co-financed by the EU and National funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH – CREATE – INNOVATE, project no. Τ2ΕΔΚ-00681

    Skip to content