Intelligent Forcipomyia taiwana monitoring and management system

Patent metadata

Field	Value
Publication number	US-12310348-B2
Application number	US-202218145015-A
Country	US
Kind code	B2
Filing date	Dec 21, 2022
Priority date	Jan 3, 2022
Publication date	May 27, 2025
Grant date	May 27, 2025

Abstract

An intelligent Forcipomyia taiwana monitoring and management system comprises: a catching mechanism grabbing a to-be-identified target; a database storing a datum comprising pictures of a flying insect category; a model training module using the pictures to establish a training model; an image capture module shooting an image including the target; an identifying module selecting a first segmented region including the target by using YOLO detection framework technology, extracting a first identification feature from the target, and inputting the feature into the training model for deep learning to identify a flying insect category to which the target belongs and produce an identification result; a counting module recording a number of the target into the database; and a predictive tracking module obtaining a marked object based on the result marked with the target identified in the image, and using a Monte-Carlo category algorithm to track and predict the object.

First claim

What is claimed is:

1. An intelligent Forcipomyia taiwana monitoring and management system comprising:
a catching mechanism having a housing, a first opening and a negative pressure device, the first opening being disposed on one side of the housing to enable an inner 5 space of the housing to communicate externally through the first opening, the negative pressure device being disposed on the housing, so that an air pressure in the inner space being lower than an external air pressure to be capable of sucking a to-be-identified target into the inner space from an external environment through the first opening, wherein the to-be-identified target is a flying insect belonging to the order Diptera of the class Insecta;
a database storing a preset datum, the datum comprising at least a predetermined number of example pictures of at least one flying insect category;
a model training module using the example pictures to perform calculations to establish a training model;
an image capture module disposed on the housing for shooting an image including the to-be-identified target;
an identifying module selecting a first segmented region including the to-be-identified target from the image by using a You Only Look Once detection framework technology, extracting at least one first identification feature from the to-be-identified target in the first segmented region, and inputting the at least one first identification feature into the training model for deep learning of image identification in order to identify a flying insect category to which the to-be-identified target belongs and producing an identification result;
a counting module recording a number of the to-be-identified target included in the identification result into the database; and
a predictive tracking module obtaining a marked object based on the identification result marked with the to-be-identified target identified in the image, and using a Monte-Carlo category algorithm to track and predict the marked object, thereby reducing a misjudgment rate in a tracking process;
wherein the catching mechanism further comprises:
a partition disposed in the housing and dividing the inner space into a first chamber and a second chamber to enable the first chamber to communicate externally through the first opening;
a through hole penetratingly disposed on the partition to enable the first chamber to communicate with the second chamber;
a second opening corresponding to a position of the second chamber and penetratingly provided on the housing to enable the second chamber to communicate externally through the second opening;
a tapered first sleeve located in the first chamber, one end opening of the first sleeve is abutted against and connected to a position of the housing corresponding to the first opening to enable the first sleeve to communicate externally through the first opening, and an inner diameter of the first sleeve gradually decreases toward a direction of the second chamber;
a tapered second sleeve located in the first chamber, one end opening of the second sleeve is abutted against and connected to a position of the partition corresponding to the through hole to enable the second sleeve to communicate with the second chamber through the through hole, and an inner diameter of the second sleeve gradually increases toward a direction of the second chamber; and
a connecting pipe bridged between the first sleeve and the second sleeve, so that the first sleeve and the second sleeve communicate with each other, and an inner diameter of the connecting pipe is equal to a minimum inner diameter of the first sleeve or equal to a minimum inner diameter of the second sleeve.

2. The intelligent Forcipomyia taiwana monitoring and management system as claimed in claim 1 , wherein a number of the image is two, and the images are arranged from front to back according to a time sequence, the predictive tracking module uses a coordinate position of the marked object in the image arranged in front as an origin, the identifying module uses the You Only Look Once detection frame technology to randomly sample a plurality of second segmented regions around a position of the origin in the image arranged at back, extracts a second identification feature from the second segmented regions respectively, compares and analyzes the second identification features to find the one with a highest degree of similarity with the at least one first identification feature, and a coordinate position of the second segmented region to which the second identification feature with the highest degree of similarity belongs is defined as a predictive position for using as a tracking prediction result of the marked object.

3. The intelligent Forcipomyia taiwana monitoring and management system as claimed in claim 1 , wherein the flying insect is a Forcipomyia taiwana or a midge or a mosquito.

4. The intelligent Forcipomyia taiwana monitoring and management system as claimed in claim 1 , wherein the catching mechanism further comprises a filter part covering the second opening and only allowing a fluid to pass through.

5. The intelligent Forcipomyia taiwana monitoring and management system as claimed in claim 1 , wherein the negative pressure device disposed on the housing is a fan located in the second chamber and is disposed on the partition corresponding to a position of the through hole.

Assignees

• Univ Feng Chia

Inventors

• Gan Yee Siang
• Fan Shih-Kai
• Tseng Ching-Yun
• Liong Sze-Teng
• Sheng yu-ting

Classifications

• G06V10/774

  Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• A01M1/08

  {Attracting and catching insects by} using combined illumination {or colours} and suction effects · CPC title

• G06T7/246

  using feature-based methods, e.g. the tracking of corners or segments · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06V40/10

  Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands · CPC title