Electronics Weekly Electronics Design & Components Tech News
“Harvesting a field as little as a day before or after the optimal time could reduce the potential income of that field for the farmer by 3.7% to as much as 20.4%,” said Tokyo researcher Wei Guo. “But optimum harvest times are not an easy thing to predict and ideally require detailed knowledge of each plant. Such data would be cost and time prohibitive if people were employed to collect it – this is where the drones come in.”
Guo’s background includes both computer science and agricultural science, and he and his team set out to analyse a field full of plants – broccoli in this case – to predict growth characteristics, using simple drones and repeated over-flights to collect image data, that was then analysed by a deep learning algorithm to catalogued the progress of every plant in the field.
The individual size of all of the vegetables in the field was predicted, while the changing market value of each size of sprout was monitored and fed into the system, allowing frequent financial forecast to be made for the farmer.
“The idea is relatively simple, but the design, implementation and execution is extraordinarily complex,” said Guo. The resulting system produces “easy-to-understand visual data for farmers [left]. Given the current relative low costs of drones and computers, a commercial version of this system should be within reach to many.”
Challenges included allowing for the way plants move in the wind and how the light changes with time and the season – requiring the team to “invest a huge amount of time labelling various aspects of images the drones might see”, according to the university. Also, the volume of information caused issues: “image data was often of the order of trillions of pixels”.
The University of Tokyo worked with Chiba University – their work is published as ‘Drone-based harvest data prediction can reduce on-farm food loss and improve farmer income’ in Plant Phenomics.
Images provided by the University of Tokyo
