Search

BR-112021012064-B1 - SYSTEM AND METHOD FOR PREDICTING INSECT ATTACK RISK

BR112021012064B1BR 112021012064 B1BR112021012064 B1BR 112021012064B1BR-112021012064-B1

Abstract

INSECT ATTACK RISK PREDICTION SYSTEM AND METHOD. An insect attack prediction system (100) is described, comprising: at least one processor provided with a plurality of software modules comprising: an insect identification module (103) configured to process at least one digital insect image (IM) to provide a presence value (IPD), representing the presence of insects in an area of interest for insect attack; a data collection module (102) configured to acquire behavioral data of insects associated with said area and comprising at least one of the following data groups: meteorological data; environmental data; historical insect presence data. The system further comprises a prediction module (104) configured to process the presence value (IPD) and the behavioral data of insects according to a mathematical prediction algorithm (302) to estimate an attack risk (PRB) for the area of interest.

Inventors

  • Andrea SOZZI SABATINI
  • Alessandra BROTZU

Assignees

  • Agrorobotica S.r.l

Dates

Publication Date
20260310
Application Date
20200206
Priority Date
20190215

Claims (9)

  1. 1. Insect attack prediction system (100), comprising: at least one processor provided with a plurality of software modules comprising: an insect identification module (103) configured to process at least one digital insect image (IM) to provide a presence value (IPD), representing the presence of insects in an area of interest for insect attack; a data collection module (102) configured to acquire behavioral data of insects associated with said area and comprising at least one of the following data groups: meteorological data; environmental data; Historical insect presence data; a forecasting module (104) configured to process the presence value (IPD) and insect behavioral data according to a mathematical forecasting algorithm (302) to estimate an attack risk (PRB) for the area of interest; characterized in that the plurality of software modules further comprises: an update and configuration module (106) configured to define a current mathematical forecasting algorithm (302) by defining a current forecasting algorithm definition dataset comprising: a forecasting model typology, algorithm configuration values, algorithm variable types; a difference matching module (113) configured to detect differences between the current forecasting algorithm definition dataset associated with the area of interest with an additional forecasting algorithm definition dataset associated with an additional area of interest or the previous acquisition time; wherein the update and configuration module (106) is further configured to update the current mathematical forecasting algorithm (302) employing the additional forecasting algorithm definition dataset.
  2. 2. System according to claim 1, characterized in that the insect identification module (103) comprises: a visual computer algorithm (300) configured to process at least one insect image and extract measured entomological parameters; an insect classification algorithm (301) configured to identify an insect from the extracted measured entomological parameters and provide the presence value.
  3. 3. System according to claim 1, characterized in that: meteorological data are selected from the following quantities: temperature, humidity, pressure, humidity level, leaf hygrometer; environmental data are selected from the following parameters: air quality, carbon dioxide concentration CO2, carbon monoxide concentration CO, volatile organic compound concentration, ammonia concentration, luminosity, presence of sound, sound level, long-term seasonal weather, presence of pesticide; historical insect presence data include data on insect attacks on the area of interest that occurred before a current time period subject to risk forecasting.
  4. 4. System according to claim 1 or 2, characterized in that the mathematical prediction algorithm (302) and the insect classification algorithm (301) can be algorithms selected from the group: neural network-based model, non-neural network-based model.
  5. 5. System according to claim 1, characterized in that the plurality of software modules further comprises: a local knowledge module (105) structured to store a set of current insect behavioral knowledge data based on a set of values assumed by at least one of the following sets: entomological parameters, meteorological quantities, environmental quantities and identified species of corresponding insects.
  6. 6. System according to claims 2 and 4, characterized in that: the said insect classification algorithm (301) is based on the current insect behavioral knowledge dataset; wherein the software plurality modulates an update module (106) configured to replace the current insect behavioral knowledge dataset being modified with an updated insect behavioral knowledge dataset.
  7. 7. System according to claim 5, characterized in that the plurality of software modules comprises a difference detection module (114) configured to: detect differences between the current insect behavioral knowledge dataset associated with the area of interest with an additional insect behavioral knowledge dataset associated with an additional area of interest or the previous acquisition time; replace the current insect behavioral knowledge dataset with the additional insect behavioral knowledge dataset in connection with said area of interest.
  8. 8. System according to claim 4, characterized in that: the non-neural network-based model is a logistic regression; the neural network-based model is selected from the group comprising: Convolutional Neural Network, Deep Neural Network.
  9. 9. Insect attack prediction method (200), comprising: processing (202) at least one digital insect image (IM) to provide a presence value (IPD), representing the presence of insects in an area of interest for insect attack; acquiring behavioral insect data associated with said area and comprising at least one of the following data groups: meteorological data; environmental data; Historical insect presence data, processing (203) the presence value (IPD) and insect behavioral data according to a mathematical prediction algorithm (302) to estimate an attack risk (PRB) for the area of interest; the insect attack prediction method (200) being characterized by further comprising: defining, by means of an update and configuration module (106), a current mathematical prediction algorithm (302), by defining a current prediction algorithm definition dataset comprising: a prediction model typology, algorithm configuration values, algorithm variable types; detecting, by means of a difference matching module (113), differences between the current prediction algorithm definition dataset associated with the area of interest with an additional prediction algorithm definition dataset associated with an additional area of interest or a previous acquisition time; updating the current mathematical prediction algorithm (302) using the additional prediction algorithm definition dataset and the aforementioned update and configuration module. (106).

Description

TECHNICAL FIELD [001] The present invention relates to a system and a method for predicting the risk of insect attack. FUNDAMENTALS OF THE INVENTION [002] In agriculture, insect attacks are one of the biggest sources of crop damage. By their nature, insect attacks appear to be unpredictable and generally different from season to season, due to a number of constantly changing factors, including, but not limited to, weather patterns, crop growth and arrangement. [003] The late discovery of insect infestation is therefore a serious problem, as available remedies may be ineffective in saving crops. In particular, the lack of efficient and timely monitoring is one of the main reasons why the use of biological pesticides is struggling to become widespread due to their limited effectiveness over time. [004] Known monitoring and forecasting techniques are based on experts (e.g., entomologists) who directly analyze crops or, according to more recent solutions, remotely assess images to identify target insects and empirically estimate possible attacks. [005] Known techniques are proving to be inefficient, as they require human intervention and skills to identify specific target insects and show significant limitations in their ability to predict insect attacks. SUMMARY OF THE INVENTION [006] The present invention addresses the problem of providing an insect attack risk prediction system that shows satisfactory prediction performance. [007] According to a first objective, the present invention relates to an insect attack prediction system defined by the attached independent claim 1. Particular embodiments of the system are described by dependent claims 2 to 9. [008] According to a second objective, the present invention refers to a method for predicting insect attacks as defined by the attached independent claim 10. BRIEF DESCRIPTION OF THE DRAWINGS [009] Other features and advantages will become more apparent from the following description of a preferred embodiment and its alternatives provided as an example with reference to the attached drawings, in which: Figure 1 schematically shows an example of an insect attack prediction system; Figure 2 is a flowchart showing an example of an insect attack prediction method, implementable by the aforementioned system. DETAILED DESCRIPTION [0010] Figure 1 schematically shows one embodiment of the insect attack prediction system 100 configured to predict the probability of an insect attack on an area of interest. The insect attack prediction system 100 (hereinafter, prediction system, for brevity) can predict insect attacks carried out, for example, by the following insect species: Bactrocera oleae, Lobesia botrana, Cydia pomonella, Cydia molesta, Cydia funebrana, Spodoptera littoralis, Spodoptera exigua, Helicoverpa armigera. The area of interest is the geographic zone, such as a particular field with different possible dimensions. The prediction system 100 can also predict attacks in more than one area of interest. [0011] In particular, the insect attack prediction system 100, as represented in Figure 1, comprises a processing device 101 and a sensing device 102 (SEN-DV). The processing device 101 comprises at least one memory configured to store data and instructions. In particular, the processing device 101 comprises at least one processor and, more particularly, it may be a network of processors (such as a telematics network) that may be organized, for example, to operate according to a cloud technology. [0012] The memories of the processing device 101 include instructions for configuring the processing device 101 in order to perform a method of predicting insect affixation. According to the example in Figure 1, some of the instructions executable by the processing device 101 can be grouped into a plurality of software modules comprising: an insect identification module 103 (INS-ID) and a risk prediction module 104 (RSK-PRD). [0013] According to the example described, the sensor device 102, the insect identification module 103 and the risk prediction module 104 are local modules, that is, they are functionally associated with a specific area of interest. If the prediction system 100 is configured to serve a plurality of different areas of interest, other sensor devices 102, other insect identification modules 103 and other risk prediction modules 104 may be employed. [0014] In addition, the following additional software modules can be run by the processor device 101: a local knowledge module (L-KW) 105 and a knowledge configuration and update module 106 (KW-CONF-UPDT). [0015] In particular, the sensing apparatus 102 comprises at least one digital camera 107 for obtaining digital images. The digital camera 107 may be a still camera or a video camera. More particularly, the digital camera 107 may be selected from the group: RGB camera, infrared camera, ultraviolet camera. [0016] In addition, the sensing apparatus 102 may comprise at least one meteorological sensor 108 config