By combining ultra-high-resolution drone footage and artificial intelligence, a new model was developed for the early detection of potato alternaria disease. According to the programmers, the neural network can also learn other diseases of other plants.
Early detection
The Alternaria fungus infects the leaves and stems and destroys the assimilation surface, causing severe yield loss. It could bring a breakthrough in precision agriculture if we are able to recognize not only nitrogen deficiency, but also similar diseases through machine “vision” and “intelligence” in time.
Belgian researchers have created an artificial neural network with patented “filters” for disease recognition that can identify and to learn the pictorial patterns characteristic of Altermária – in this case, the small spots on infrared images – similarly to how the human brain learns from experience. This allows the model to recognize the disease before it becomes obvious to humans.
Targeted treatment
The team trained the model on data sets collected over four growing seasons, and it now works in conditions it wasn’t “trained” for. Thanks to technology, the boards – similarly to the yield maps – “disease map” can be drawn, on the basis of which application commands, i.e. application maps, can be created for the sprayer. In Hungarian, it becomes possible to spot treatment in the stock. Fewer released chemicals result in cost savings on the one hand, and environmental protection on the other.
The research team is optimistic about the future applications of their work. They believe that the neural model will not only be able to identify potato alternaria disease, but will also be able to learn other diseases of other plants.
You may also be interested in this article: Robotic weeding in ornamental plants.
Tags: involved identification plant diseases
