Lecture 1: Non-invasive optical sensing approaches to seed phenotyping : opportunities and challenges
By Puneet Mishra – Wageningen University
This presentation will provide a general overview of the current state-of-the-art of non-invasive sensing technologies to support seed phenotyping. The particular focus will be optical sensing techniques which means studying the interaction of light with the seeds. In recent years, major developments have taken place in both the hardware and software, and several optical sensing techniques such as colour imaging, 3D volume imaging, multispectral, hyperspectral and X-ray imaging, have become popular tools to explore the physicochemical properties of seeds. Although some techniques are now regularly used for see phenotyping while some are still at the research forefront. The main benefit of optical techniques is that they allow a high-throughput analysis of seeds, while minimally influencing them. However, optical sensing techniques also have their limitations, which the users should be aware of to reap the most benefit from them.
Lecture 2: Seed phenotyping with multi-modal imaging and AI analysis
By Fred Hugen – SeQso
In the past, various imaging techniques have been developed and used for seed analysis. This presentation is about an integrated approach to combine multiple imaging technologies for seed phenotyping. Yielding a rich data-set for each single seed. Thanks to current developments in digital phenotyping of plants, large labelled datasets can be produced. Enabling the use for applying modern AI technologies to discover correlations between “multi-modal” phenotype of seeds and specific characteristics/traits of seeds.
Lecture 3: Digital seeds: Does your digital seed germinate?
By Jens Michael Carstensen – Videometer
This presentation explores the use of rich seed databases based on high quality spectral images of seeds. These digital twins of seeds represent an effective and efficient way to preserve seed properties while the physical seed samples changes over time. The digital seeds may be analyzed for purity, damages, health, germination, vigor, longevity, and many other traits, and population studies with many seeds, e.g. 100,000 seeds, can be performed in order to better predict field or production performance. For seed testing then seed databases allows not only to accumulate and maintain phenotype knowledge, but also automatic update and signing of classifiers as well as virtual proficiency tests among classifiers. All together this allows for seed analysis and seed testing going digital.