Artificial intelligence and image processing-based portable pest detection device and working method

Abstract

The invention relates to an artificial intelligence and image processing-based portable pest detection device and working method that aims to increase efficiency in agriculture by minimising losses caused by pests in agricultural crops and by providing fast and accurate damage detection through high-resolution imaging, strong hardware support, and artificial intelligence-based analysis features. The invention relates to an artificial intelligence and image processing-based portable pest detection device and working method which provides a portable image processing system for the detection and analysis of damage caused by pest organisms (mites and insects) in agricultural crops, for the purpose of increasing efficiency in agricultural production, and which comprises a processor, a high-resolution CSI camera module, an LED display module, a battery module, and a memory card.

Description

[0001] ARTIFICIAL INTELLIGENCE AND IMAGE PROCESSING-BASED PORTABLE PEST DETECTION DEVICE AND WORKING METHOD

[0002] Technical Field of the Invention:

[0003] The invention relates to an artificial intelligence and image processing-based portable pest detection device and working method that aims to increase efficiency in agriculture by minimising losses caused by pests in agricultural crops and by providing fast and accurate damage detection through high-resolution imaging, strong hardware support, and artificial intelligence-based analysis features.

[0004] The invention relates to an artificial intelligence and image processing-based portable pest detection device and working method which provides a portable image processing system for the detection and analysis of damage caused by pest organisms (mites and insects) in agricultural crops, for the purpose of increasing efficiency in agricultural production, and which comprises a processor, a high-resolution CSI camera module, an LED display module, a battery module, and a memory card.

[0005] State of the Art:

[0006] Technological developments in the field of pest detection and plant health monitoring in agriculture are increasingly based on artificial intelligence (Al) and image processing methods. These technologies have great potential for increasing efficiency in agriculture and for preventing the spread of diseases. Image processing- and Al-based systems are capable of detecting agricultural pests quickly and accurately, enabling early intervention by providing farmers with real-time data.

[0007] Applications, products, patents, and academic publications developed to date demonstrate that significant progress has been achieved in this field. Methods used for pest detection and plant health monitoring in agriculture are generally based on technologies such as high-resolution cameras or drones. These technologies use object detection algorithms (such as YOLO, Fast R-CNN, and SSD) to capture andanalyse images of plants. In addition, methods for monitoring plant stress and detecting diseases through spectral analysis techniques are also widely used.

[0008] These advanced systems operate with trained models to identify pests and diseases on plant leaves. In particular, while plant health is monitored using methods such as NDVI (Normalized Difference Vegetation Index), images of plants in various spectra are analysed using multispectral cameras. Such systems help prevent agricultural damage by detecting pests in a timely manner.

[0009] Among similar products used in agriculture, platforms such as FieldView and Taranis stand out. FieldView provides field monitoring through drones and satellite images and offers Al-based solutions for plant health monitoring and pest detection. Taranis, on the other hand, facilitates disease detection using high-resolution cameras. Other products such as Precision Hawk and Slantrange provide drone-based multispectral imaging solutions. These systems are particularly preferred by large-scale agricultural enterprises and consultancy companies.

[0010] Although various proposals and applications have been developed for pest detection devices in the state of the art, these developments are not sufficient. For this purpose, some applications relating to inventions developed are presented below.

[0011] The patent application numbered “CN113780357A” in the state of the art aims to develop an artificial intelligence (Al)-based method for identifying com leaf diseases and insect pests via mobile terminal devices. The method is designed to provide fast and accurate image processing on mobile devices by using MobileNet convolutional neural networks (CNN) and transfer learning techniques. This process includes the steps of preparing a dataset, processing the data, creating a model, training, and optimisation. Finally, the trained model is converted into TFIite format and embedded into a mobile terminal application, enabling the identification of com leaf diseases and pests via the devices. This approach enables high-precision scanning on mobile devices with low operating costs and high speed.

[0012] The patent application numbered “US2014311014A1” in the state of the art relates to a pest control system capable of detecting and destroying agricultural pests without causing harm to crops and other living organisms. The system comprises a detection beam emitter for detecting pests, a detection tool that detects pests by receiving signals reflected by the detection beam, a targeting tool that directs a destruction beamtowards the detected pests to eliminate them, and a destruction beam emitter. This method enables effective pest control without the use of chemical pesticides.

[0013] The patent application numbered “CN118570510A” in the state of the art describes a method and a device related to the detection of agricultural pests and diseases. By using an artificial intelligence language model, it is aimed to automatically analyse and detect the visual characteristics of pest and disease damage in a specific crop. In this process, the visual characteristics of pests and diseases are stored in a pre-established library, and the crop image is analysed against this library to determine the type of damage. The artificial intelligence provides accurate diagnosis by performing scoring in order to correctly detect pests and diseases in the crop.

[0014] In the state of the art, there is a need for an artificial intelligence and image processingbased portable pest detection device and working method that aims to increase efficiency in agriculture by minimising losses caused by pests in agricultural crops and by providing fast and accurate damage detection through high-resolution imaging, strong hardware support, and artificial intelligence-based analysis features.

[0015] As a result, due to the disadvantages described above and the inadequacy of existing solutions with respect to the subject matter, it has become necessary to carry out a development in the relevant technical field.

[0016] The Aim of the Invention:

[0017] The most important aim of the invention is to enable the rapid, accurate, and effective detection of damage caused by pest organisms (mites and insects) in agricultural crops.

[0018] Another aim of the invention is to easily detect pests and to determine plant diseases with high accuracy by combining high-resolution image processing techniques and artificial intelligence-based analysis methods.

[0019] Another aim of the invention is to provide a more efficient and practical agricultural pest detection solution compared to existing fixed systems by offering a portable and integrable solution.Another aim of the invention is to sensitively detect small pests and disease symptoms by obtaining detailed images through 4-8K resolution CSI camera modules. Thus, it enables diseased regions to be clearly identified and ensures that better results are obtained over time through continuous learning capability.

[0020] Another aim of the invention is to enable real-time analysis of high-resolution images by providing strong processor support and to accelerate data collection and analysis processes by minimising farmers’ loss of time.

[0021] Another aim of the invention is to ensure applicability across different agricultural crops through its flexible structure, to increase sustainability in agricultural production, and to perform pest control more effectively.

[0022] Brief Description of the Drawings:

[0023] FIGURE-1 is a drawing showing the connection diagram of the artificial intelligence and image processing-based portable pest detection device system that is the subject of the invention.

[0024] FIGURE-2 is a drawing showing the image of the artificial intelligence and image processing-based portable pest detection device that is the subject of the invention.

[0025] Reference Numbers:

[0026] 1: Processor

[0027] 2: CSI camera module

[0028] 3: LED display module

[0029] 4: Battery module

[0030] 5: Memory card

[0031] 7: Portable pest detection deviceDescription of the Invention:

[0032] In general terms, the invention is an artificial intelligence and image processing-based portable pest detection device (7) and working method which enable pest detection in agricultural crops with high accuracy, analyse the spread of damage, and provide farmers with an effective opportunity for intervention, and which generally comprise a processor (1), a high-resolution CSI camera module (2), an LED display module (3), a battery module (4), and a memory card (5), and which enable the detection of damage caused by mites and insects.

[0033] The invention enables the rapid, accurate, and effective detection of damage caused by pest organisms (mites and insects) in agricultural crops. By combining high-resolution image processing techniques and artificial intelligence-based analysis methods, it enables pests to be easily detected and plant diseases to be determined with high accuracy.

[0034] The invention provides a more efficient and practical agricultural pest detection solution compared to existing fixed systems by offering a portable and integrable solution. By obtaining detailed images through 4-8K resolution CSI camera modules (2), the invention sensitively detects small pests and disease symptoms. Thus, it enables diseased regions to be clearly identified and ensures that better results are obtained over time through continuous learning capability.

[0035] By providing strong processor (1) support, the invention enables real-time analysis of high-resolution images and accelerates data collection and analysis processes by minimising farmers’ loss of time. Owing to its flexible structure, it ensures applicability across different agricultural crops, increases sustainability in agricultural production, and performs pest control more effectively.

[0036] The processor (1) is an element used as the central unit of the system for performing artificial intelligence and image processing tasks, offering high parallel processing capacity and energy efficiency, and, by means of CLIDA cores, rapidly running deep learning models to instantly analyse large datasets.

[0037] The high-resolution CSI camera module (2) is a Lucid Vision Phoenix 8K or similar CSI camera module (2) which captures detailed images of agricultural crops by means of micro-lens technology, enables the detection of small pests and symptoms at the initialstages of damage, and allows images to be captured from various angles and under different lighting conditions.

[0038] The LED display module (3) is an element connected to the processor (1 ) and provides information by visually presenting the obtained analysis results, damaged regions, damage spread rate, and confidence scores to the user.

[0039] The battery module (4) is a structure that supplies the processor (1) and other hardware components with 5V-4A power. The memory card (5) is a 64 GB structure in which analysis results and data are stored. This structure increases the portability of the system and its data storage capacity. The YOLOv8x-seg model is used as the object detection model.

[0040] The working method of the artificial intelligence and image processing-based portable pest detection device comprises the process steps of;

[0041] • Activation of the interface software loaded on the memory card upon initiation of the system,

[0042] • The interface software communicating with the CSI camera module connected to the system,

[0043] • High-resolution images being captured from the environment via the CSI camera module in order to enrich the dataset from different angles and under different lighting conditions,

[0044] • Applying processes such as brightness enhancement, blurring, and noise addition to the images acquired through image processing,

[0045] • The images being transmitted to the YOLOv8x-seg model stored on the memory card,

[0046] • Determining pest organisms in agricultural crops by performing segmentation on the image using the YOLOv8x-seg model,

[0047] • Identifying the living species causing damage to plants using the YOLOv8x-seg model,

[0048] • Calculating the damaged regions and the damage rate using the YOLOv8x-seg model,

[0049] • Masking the damaged regions detected by the YOLOv8x-seg model in white and the other regions in black,• Calculating the spread rate of the detected damage as a percentage via the processor,

[0050] • Calculating the confidence score via the processor, and

[0051] • Visually displaying detailed reports and analysis results regarding the damage in the agricultural area and the detected pest species to the user via the LED display module connected to the processor.

[0052] When it is necessary to proceed to a new calculation process after an operation is performed, the previously obtained data are deleted from the system’s memory unit, and the system is made ready for a new processing cycle. When a new image is captured via the interface, the same analysis process is repeated, and the system continues to maintain this cycle.

[0053] The invention can be easily transported and used in different areas under field conditions. The compact structure of the Jetson Xavier AGX processor (1) and its support by the 5V-4A battery module (4) enable the invention to be used in a mobile manner. This feature enables farmers to perform rapid detection in different fields and over large agricultural areas.

[0054] Within the scope of the invention, a Lucid Vision Phoenix 8K or similar high-resolution CSI camera module (2) is provided. This module acquires images at a resolution of 4-8K, thereby enabling detailed analysis of small pests and diseased regions. Its ability to capture images from various angles and under different lighting conditions increases image accuracy and allows more sensitive detection.

[0055] The processor (1), which is a Jetson Xavier AGX 16GB platform, constitutes the core of the system. By means of CUDA cores, deep learning algorithms are run rapidly and efficiently. The Jetson Xavier AGX processor (1) is optimised for artificial intelligence and image processing tasks and offers the capability to instantly analyse large datasets with high parallel processing capacity.

[0056] The invention identifies plant diseases with high accuracy by using the Y0L0v8x-seg model and clearly determines diseased regions through its segmentation capabilities. This model automatically detects pests and provides real-time feedback to the user. Its continuous learning capability ensures that the model produces more accurate results over time.The invention provides real-time feedback and visualisation. Analysis results are instantly transmitted to the user via an 8-inch LED display module (3). While providing visual feedback in the field, this LED display module (3) presents metrics such as damaged regions, spread rate, and confidence scores in detail. Users can make immediate interventions based on the condition of the pests by using these data. Within the scope of the invention, a memory card (5) having a capacity of 64 GB is used for storing the analysed data and images. The memory card (5) enables large datasets to be stored locally and provides easy access to the data in the future. The data storage feature allows users to re-access analysis results and to evaluate these results by making comparisons.

[0057] Within the scope of the invention, energy efficiency is prioritised in order to support long-term use in agricultural areas. While the 5V-4A battery module (4) supplies sufficient energy to all components, the invention can be operated in a portable structure by means of the optimised power consumption of the Jetson Xavier AGX processor (1).

[0058] The invention meets needs such as high accuracy, portability, strong processing capacity, and real-time feedback in agricultural areas. In this way, users can obtain faster, more reliable, and more efficient solutions in pest detection and damage analysis.

Claims

CLAIMS1. An artificial intelligence and image processing-based portable pest detection device (7), comprising:• a processor (1) used as a central unit of the system for performing artificial intelligence and image processing tasks and capable of instantly analysing large datasets;• a high-resolution CSI camera module (2) configured to acquire detailed images of agricultural crops, to detect small pests and symptoms of damage, and to enable images to be captured from various angles and under different lighting conditions;• an LED display module (3) connected to the processor and configured to visually present to a user the obtained analysis results, damaged regions, damage spread rate, and confidence scores;• a battery module (4) configured to enable operation of the processor, the CSI camera module, the LED display module, and the memory card; and • a memory card (5) in which analysis results and data are stored.

2. A working method of the artificial intelligence and image processing-based portable pest detection device (7), comprising the process steps of:• activating interface software loaded on a memory card upon initiation of the system;• the interface software communicating with a CSI camera module connected to the system;• capturing high-resolution images from the environment via the CSI camera module in order to enrich a dataset from different angles and under different lighting conditions;• applying processes such as brightness enhancement, blurring, and noise addition to the images acquired through image processing;• transmitting the images to an object detection model stored on the memory card;• determining pest organisms in agricultural crops by performing segmentation on the image using the object detection model;• identifying living species causing damage to plants using the object detection model;• calculating damaged regions and a damage rate using the object detection model;• masking damaged regions detected by the object detection model in white and other regions in black;• calculating a damage spread rate as a percentage via a processor;• calculating a confidence score via the processor; and• visually displaying, to a user via an LED display module connected to the processor, detailed reports and analysis results regarding damage in an agricultural area and detected pest species.

3. A working method of the artificial intelligence and image processing-based portable pest detection device (7) according to claim 2, wherein, in the step of determining pest organisms in agricultural crops by performing segmentation on an image using an object detection model, said pest organisms are mites and insects.

4. A working method of the artificial intelligence and image processing-based portable pest detection device (7) according to claim 2, wherein, in the steps of transmitting images to an object detection model stored on a memory card, determining pest organisms in agricultural crops by performing segmentation on an image, identifying living species causing damage to plants, calculating damaged regions and a damage rate, and masking damaged regions detected by the object detection model in white and other regions in black, the object detection model used is a Y0L0v8x-seg model.

Images