Spectro-AG B.V. focuses on the design and development of hyperspectral and multispectral sensors for Unmanned Aerial Vehicle (UAV) and Unmanned Ground Vehicle (UGV) systems. We leverage sensors and artificial intelligence, including machine learning and deep learning algorithms, to generate new knowledge and solutions for precision agriculture and environmental protection. HyperSlit is a target-focused hyperspectral sensor with 1000 spectral bands. With its two hyperspectral sensors for sensing targets and light source irradiance, the compact and drone ready HyperSlit provides a unique opportunity to create geotagged spectral libraries. The operation of HyperSlit is automatic and requires minimal human intervention in terms of corrections and calibrations.
We also offer Geo-AI solutions for calibration and processing multispectral and hyperspectral satellite image. For example, Deep4Sat which creates and operationalizes deep learning models for object recognition from space. Users can use Deep4Sat web-application to train an AI model to detect and localize their target objects in high resolution (30-50cm) hyperspectral and multispectral satellite images. The application allows users to label their individual objects of interest on their satellite images for training Deep Learning models. The trained deep Convolutional Neural Network (CNN) models detect the objects in the satellite images and their geographic locations are reported to the users.
In Spectro-AG we aim on research and development of solution products that support decision makers, with accurate spatio-spectral information. For that reason, HyperSlit can be used as a hyperspectral and multispectral satellite image calibration tool before they are imported in Deep4Sat for object recognition.