|
India faces tremendous societal and ecological challenges. Cities are growing which is accompanied by an increase in population and consequently traffic. At the same time, while India's forest cover is on average increasing, it is not clear how much of this is due to plantation in contrast to natural forest, a knowledge gap that is possibly endangering biodiversity of India's forests.
Standardly collected remote sensing data of India offers a great opportunity for quantifying the status quo of these factors and turning them into ecological and health models that can inform new government policies to help tackle these challenges.
In this talk we discuss novel image analysis and machine learning methods for analysing image data obtained from satellites and more localised on the ground imaging such as traffic cameras. Our main objectives are traffic quantification and traffic mode classification, and mapping tree species. Our proposed methods are aimed to overcoming the lack of existing annotated data as well as the multi-scale character of the measurements, fusing information from satellite images with traffic camera data.
|
