In order to reduce the use of herbicides, there are basically 2 options:
- How to get rid of the plant between the rows?
- How to get rid of the plant on the row?
What remains an issue is how to replace the selectivity of chemical rather finding tools to remove the weed.
The deep learning methods seem to be a good response, especially the deep Neural Nets that learn the features of the crops themselves, with a huge amount of data. Deepfield Robotics now has 12 Million images, the biggest plant database in the world. They use this information for the mechanical tool, that goes in and out of the row to remove the weeds, or it can be a spot sprayer that isolates specifically the weed for the chemical spray.
Deepfield Robotics supports the integration of its system for its customers and partners.
Maurice Gohlke explains everything.