null
 

Computer vision in citrus

image annotation in citrus tree for pest detection

Computer vision can be fundamental in agriculture. It plays a key role in crop resistance against pests, diseases and nutrient deficiencies and provides a method of early prediction and control of plant enemies. Accurate detection of plant enemies leads to increase of productivity, prevents quality loss and decreases the use of plant protection products (PPPs). Citrus are quite vulnerable and sensitive to many pests and diseases and very demanding in fertile soils. So, it is necessary for citrus producers to take care of their yield and to have a guaranteed way to protect their crops. This very demanding issue can be worked out by using Artificial Intelligence (AI) techniques and machine vision devices in order to scan the field for any presence of infestation.

Citrus trees are a target for many pests and diseases almost all year round because citrus trees thrive in tropical and subtropical zones of the planet which means that warm temperature and high humidity enhances the occurrence of spread of the diseases. Citrus trees have huge and high dense canopy and because of that many pest and disease infestations hide in the back of the leaves, in the young vegetation, even in cavities in the wood. As far as the nutrient deficiencies are concerned, these become visible when the losses are significant and usually happen early in the spring because of the high demands during that period. Furthermore, different nutrient deficiencies may have the same indication especially when emerging. Here come computer vision devices which use algorithms to detect the presence of pest infestations, diseases and nutrient deficiencies. The algorithms have been trained with real data to recognize the most difficult cases.

Computer vision already struggles with the most difficult cases of citrus tree infestations with successful results. Undoubtedly, computer vision achieves exceptional accuracy in prediction due to trained neural networks and the computer’s ability not to get tired over the time, as happens in humans. One more positive aspect is the huge variety of computer vision systems and devices which someone can find. A device can be portable or can be mounted in a tractor or drone. It can also be implemented as a mobile application using the phone’s camera. In any case, it is very easy to use and really effective. One more advantage of computer vision in agriculture is that many tiring and time-consuming tasks can be eliminated. For example, producers must inspect their orchards very often to see if everything is fine and may need to do this more frequently if the weather conditions are suitable for pests’ and diseases’ spread. This is tedious even for only a few trees. What if the producer has 100 or 1000 hectares (ha) of citrus trees? That is where computer vision is needed. It should not be omitted the significant contribution of the computer vision in the maturity estimation of the citrus fruits which supports producer’s decision making for the best quality level of the harvested fruits.

Citrus trees and their products are valuable for human health and must be treated in the best way. Therefore, agricultural engineering makes every possible effort to maximize the production of citrus trees while maintaining the quality levels high, supporting the producer and protecting the environment.

References:

  1. https://www.europarl.europa.eu/RegData/etudes/IDAN/2019/634416/EPRS_IDA(2019)634416_EN.pdf
  2.  https://arxiv.org/ftp/arxiv/papers/1807/1807.11809.pdf
  3. https://www.fao.org/3/ac681e/ac681e08.htm
  4. https://ijabe.org/index.php/ijabe/article/view/5607/pdf