The partners want to collaborate on product/prototype development for computer vision-based solutions for the rapidly growing greenhouse sector to enable operational optimization.
This will enable greenhouse owners to automate, via mobile cameras, the counting and measure of fruit and vegetables in the greenhouse, which are currently two of the most time-consuming and complicated tasks in greenhouses. In extensive ‘client discovery’ with greenhouses in the Netherlands and Germany we established that the demand for automation is high. This can only be done via ‘Computer Vision’ / CV (in essence, artificial intelligence applied to the recognition of fruit and vegetables), which in turn needs a “mobile component” so that the greenhouse does not have to use too many cameras.
In addition, the project partners use computer vision to understand the rate of growth of fruit / vegetables in the greenhouse, thus enabling greenhouse owners to predict and even influence when they can harvest how much by adjusting relevant inputs. This addresses important pain points in the greenhouse industry such as reducing the current significant loss of produce due to misalignment between the timing of harvest and sale; or enabling greenhouse contract to make more accurate commitments vis-a-vis buyers such as supermarket chains about when how the greenhouse will deliver how much produce.
The project is structured into the following 3 work packages (WP):
Firstly, the development of camera mobility solutions. These consist of two ‘autonomous’ mobility solutions, namely drones and robots; and two non-autonomous solutions, namely cell phone app camera-based imaging and small ‘towers’ of cameras that can be integraated in existing transport solutions in greenhouses.
Secondly, transferability of the computer vision work to various crops.
Thirdly, linking CV-based prediction with price prediction to enable greenhouses to predict and plan their future revenue.
