In the past, significant improvements in yield potential have been made through optimizing harvest index. However, gains in yield potential and yield stability in the future will also rely on improved resource use efficiency (nutrients, water, and light) and optimized shoot growth. This will require an understanding of shoot growth, how it is controlled and how environmental conditions affect growth.

If we want to identify the underlying genetics of shoot growth, we require suitable phenotyping methods at a throughput that would allow a forward genetics approach. While phenotyping methods to study total shoot growth have become available for wheat, particularly in controlled environments, a dissection of the shoot into its components has been a limiting factor. While robust image analysis methods exist for plant species with a simpler 30 structure like maize, wheat and barley are more challenging. Thin, flexible leaves that twist along their axis make image based 30 modelling particularly difficult, even in controlled environment experiments. Previous methods attempting 30 modelling in small grain cereals have used hundreds of images. Here, we present a novel approach using only five side view images to ensure high throughput is possible. Our algorithm generates a large number of potential 30 models based on a small set of images and then selects the most likely model with the least number of images. We have validated the method against manual measurements of individual leaf length in wheat and barley. We have further shown that this approach can be used at high throughput on hundreds of plants over time.