Automated pattern recognition in crop plant phenotyping
Fraunhofer Institute IFF Magdeburg, Biosystems Engineering
Breeding of novel crop plants has become increasingly important to
address the challenges of growing earth population, climate change,
and demand of regenerative energy (e.g. bio-fuels) as well as mineral
oil substitutes (bio-polymers).
This requires crop plants with tailored and very specific
properties. Regardless whether these plants are developed by
classical plant breeding or direct biomolecular manipulation
(GM plants), a statistically valid and high-throughput
assessment of the plant's phenotype is always essential.
In order to assess the relevant attributes of a plant's phenotype,
typically the recognition and modelling of spatiotemporal development
patterns needs to be done.
By means of a number of challenging real-world applications this talk
will demonstrate how computational intelligence and machine learning
assists this endeavour.