KR20190063184A - Detection of dangerous things to infants through image analysis and deep learning - Google Patents

Abstract

According to one embodiment of the present invention, provided is a method for detecting a dangerous object for infants and young children using an image analysis and deep learning, which comprises the steps of: using a deep learning algorithm from an image including a plurality of peripheral objects infants and young children to recognize the plurality of peripheral objects; analyzing coordinate information on the peripheral objects recognized in the image; and predicting a dangerous situation of the infants and the young children based on the analysis.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting dangerous substances in infants and young children using image analysis and deep running,

More particularly, the present invention relates to a method for detecting a dangerous object in an infant using a deep learning algorithm capable of recognizing various objects.

Because of the nature of young children, it is possible to expose them to dangerous situations even if they sell their eyes for a while. In order to prevent this, there are many cases of object recognition using object recognition and image processing algorithms based on various sensors.

However, in the sensor-based case, there is a problem that the sensor is always attached to the body in order to obtain the biometric information and the external environment information, and a high cost sensor network is required for reliable data collection. In the image processing algorithm, there are many computation steps for image correction and it is difficult to apply one algorithm to many objects, so there is a limitation in applying them to various objects.

A related prior art is Korean Patent Registration No. 10-0685134 entitled " Camera based monitoring method and system, public announcement date: February 22, 2007 ".

In one embodiment of the present invention, an image analysis is performed to detect various objects around infants and children by recognizing surrounding objects using a deep learning algorithm from images including infants and surrounding objects, and analyzing coordinate information on surrounding objects in the images. And a method for detecting dangerous infants and young children using deep running.

The problems to be solved by the present invention are not limited to the above-mentioned problem (s), and another problem (s) not mentioned can be clearly understood by those skilled in the art from the following description.

A method for detecting dangerous infants and infants using image analysis and deep learning according to an embodiment of the present invention includes the steps of recognizing the plurality of surrounding objects from an image including an infant and a plurality of surrounding objects using a deep learning algorithm, Analyzing coordinate information on the perceived plurality of peripheral objects in the at least one child, and predicting the risk situation of the infant and child based on the analysis.

The details of other embodiments are included in the detailed description and the accompanying drawings.

According to an embodiment of the present invention, a surrounding object is recognized using a deep learning algorithm from an image including infants and surrounding objects, and coordinate information on objects surrounding the image is analyzed to detect various objects around infants and young children have.

FIG. 1 is a view for explaining an infant and child dangerous material detection method using image analysis and deep learning according to an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The advantages and / or features of the present invention, and how to accomplish them, will become apparent with reference to the embodiments described in detail below with reference to the accompanying drawings. It should be understood, however, that the invention is not limited to the disclosed embodiments, but is capable of many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, To fully disclose the scope of the invention to those skilled in the art, and the invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a flowchart illustrating a method for detecting dangerous infants and children using image analysis and deep learning according to an exemplary embodiment of the present invention. Referring to FIG. 1, Analyzing coordinate information for the perceived plurality of nearby objects within the image, and predicting the risk situation of the infant based on the analysis.

The characteristics of the method for detecting dangerous infants and toddlers according to the present invention are as follows.

1. DeepLearning Preprocessor

- Collecting infant and image data through Crowling

- Image labeling preprocessing technology to construct datasets

- Use initial values to reduce the number of repetitions of training and improve accuracy

2. Neural Network Configuration

- Artificial neural network configuration using TensorFlow and high performance GPU

- Training through the Faster R-CNN, a real-time predictable model.

- Use initial values to reduce the number of repetitions of training and improve accuracy

3. Risk prediction model configuration

- Establishing a situation judgment program by setting the risk level based on recognized objects

- Determination of danger situation of infants through distance factor using object coordinates

- Logical visualization model

Specific features of the method for detecting dangerous substances for infants and children of the present invention are as follows.

S1-1. A technology that recognizes dangerous objects to infants using Faster R-CNN, a deep-running algorithm.

S1-2. A system for predicting risk situations based on recognized objects.

S2-1. In S1-1, a data set preprocessing technique for neural network training.

S2-2. In S1-1, pre-learning techniques are applied to improve the number of repetitions and accuracy of training.

S2-3. In S1-2, a system for predicting a risk situation based on the presence or absence of adults and the type of dangerous goods.

S3-1. In S2-1, the coordinates of the type and position of the internal object in the image of the data set are generated as a JSON format file.

S3-2. In S2-2, a system is used to generate a pre-learning dataset and use it as an initial value in the first training of the neural network.

S3-3. In S2-3, the distance between infants and objects is relatively determined using the kind of object and the recognized coordinate information, and the system is utilized for prediction.

S4-1. In S3-1, a system that makes the labeling result into a dataset in the tensor form of Tensorflow.

S4-2. In S3-3, a real-time log visualization program is built based on the predicted information.

S5-1. In S4-1, the system can be used not only as an image but also as an input for a camera, a CCTV, and the like.

According to the features of the present invention described above, it is possible to apply the present invention to a system for detecting real-time at a time of about 0.2 seconds per image in predicting a dangerous situation, and to build a system based on an image that can be easily collected in real life It can be used as an input of cameras, CCTV, etc.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the scope of the appended claims and equivalents thereof.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, Modification is possible. Accordingly, the spirit of the present invention should be understood only in accordance with the following claims, and all equivalents or equivalent variations thereof are included in the scope of the present invention.

Claims (1)

Recognizing the plurality of surrounding objects from the image including the infant and the plurality of surrounding objects using a deep learning algorithm;
Analyzing coordinate information for the perceived plurality of nearby objects within the image; And
Predicting the risk situation of the infants based on the analysis,
And analyzing the image of the dangerous object of infant and child by using image analysis and deep running.

Images