3D reconstruction of plant structures using AI

3D reconstruction from images of different angles has been thoughroughly studied. However, it was difficult to reconstruct the parts of objects which have hidden portions, such as plants with branches hidden under their leaves.



By combining the original image-to-image approach in a Bayesian deep learning framework and 3D reconstruction, a group of researchers led by Fumio Okura were able to estimate the location of branches that are hidden under leaves in images. Using these estimated branch positions, they achieved 3D reconstruction of plant structure.

The results are incredibly important for cultivation and management of plants by artificial intelligence. They will be greatly helpful for future cultivation technology as it allows for detailed management of growing plants in places where cultivators' can't see, coming up with best trimming or pruning methods for plants, and predicting future growth of plants.

https://www.sciencedaily.com/releases/2018/05/180507090257.htm