Search

KR-102963703-B1 - Picture Object Recognition Method Using Deep Learning Model And Computer-Readable Recording Medium Recording A Program That Performs The Same

KR102963703B1KR 102963703 B1KR102963703 B1KR 102963703B1KR-102963703-B1

Abstract

The present invention relates to a method for recognizing picture objects using a deep learning model and a computer-readable recording medium storing a program for performing the same. More specifically, the invention relates to a method for recognizing picture objects using a deep learning model and a computer-readable recording medium storing a program for performing the same, comprising: a learning dataset acquisition step in which a learning dataset including a plurality of PITR learning images of a person in the rain (PITR) is acquired; a model learning step in which a deep learning model is trained using the learning dataset; an image acquisition step in which a PITR subject image of a person in the rain (PITR) is acquired; and an object recognition step in which at least one of a stress scale and a coping resource scale within the PITR subject image is recognized as an object by inputting the PITR subject image into the deep learning model trained from the model learning step.

Inventors

  • 차유정
  • 임예원
  • 전우진

Assignees

  • 고신대학교 산학협력단

Dates

Publication Date
20260511
Application Date
20230315

Claims (5)

  1. A training dataset acquisition step in which a training dataset including a PITR training image depicting a Person in the Rain (PITR) is acquired by at least one processor; A model training step in which a deep learning model is trained using the training dataset by the above at least one processor; An image acquisition step in which a PITR subject image in which the subject draws a Person in the Rain (PITR) is acquired by the above-mentioned at least one processor; and The method comprises an object recognition step in which at least one of a stress scale and a coping resource scale within the PITR subject image is recognized as an object by inputting the PITR subject image into a deep learning model learned from the model learning step by the above-mentioned at least one processor, and The above object recognition step is, A feature map extraction step in which a hierarchical feature map for the PITR subject image is extracted by utilizing a backbone network within the deep learning model; An anchor box extraction step in which an anchor box for an object within the PITR subject image is extracted using a Feature Pyramid Network (FPN) within the deep learning model; An IoU calculation step in which, when multiple anchor boxes are extracted for the same object within the above PITR subject image, an IoU value for each of the multiple anchor boxes is calculated; A reliability calculation step in which, when each IoU value exceeds a threshold, the following [Equation 1] is used to calculate the reliability for each IoU value, and the anchor box with the lowest calculated reliability is maintained; and A picture object recognition method using a deep learning model, characterized by including an object detection step in which an object detector within the deep learning model is used to extract a class label and a bounding box for an anchor box of an object within the PITR subject image. [Mathematical Formula 1] Here, f(IoU) is the confidence level of an arbitrary anchor box, and IoU is the IoU value calculated for an arbitrary anchor box from the IoU calculation step.
  2. A computer-readable recording medium storing a program that performs a picture object recognition method using the deep learning model of claim 1.
  3. delete
  4. delete
  5. delete

Description

Picture Object Recognition Method Using Deep Learning Model And Computer-Readable Recording Medium Recording A Program That Performs The Same The present invention relates to a method for recognizing picture objects using a deep learning model that applies the field of computer vision to a PITR (Person in the rain) test, and a computer-readable recording medium that records a program for performing the same. Computer vision is a field of artificial intelligence that uses computers to replicate general human visual perception capabilities. This field of computer vision is being applied in various sectors, including medicine. The PITR (Person in the Rain) test is a psychological drawing test developed by adding a rain scene to a figure drawing. It is a test that has been verified through clinical cases to be valuable as a diagnostic tool for determining the strength of the ego, as well as for assessing stress and verifying the adequacy of coping resources. The administration method involves providing instructions to draw a person in the rain while it is raining. After the subject completes the drawing, the examiner asks questions about the order in which the drawing was made, the figure within the picture, and the actions the figure performs. Subsequently, the examiner and the subject engage in a conversation related to the drawing. Based on interpretation methods and clinical experience, the examiner may interpret the drawing by focusing on the person, rain, clouds, puddles, lightning, and the form of the rain. However, as previously mentioned, conventional methods are interpreted based on the examiner's clinical experience, meaning that interpretation methods can vary depending on the examiner's know-how. Furthermore, while the purpose is to diagnose psychological states, test subjects are prone to concealing their emotions during conversation and may intentionally refrain from speaking about difficult topics. Additionally, since a long conversation between the examiner and the test subject is mandatory, there are limitations in obtaining appropriate test results when dealing with subjects who find it difficult to engage in prolonged conversation, such as children, people with disabilities, and the elderly. Therefore, there is an urgent need for technology in this field to resolve these issues. Figure 1 is a flowchart of a picture object recognition method using a deep learning model of the present invention. FIG. 2 is a structural diagram of a deep learning model according to an embodiment of the present invention. Figure 3 is an image of a PITR subject with a class label and bounding box for a non-object, which is a stress scale, displayed according to an embodiment of the present invention. The terms used in this specification have been selected based on currently widely used general terms whenever possible, taking into account their functions in the present invention; however, these terms may vary depending on the intent of those skilled in the art, case law, the emergence of new technologies, etc. Additionally, in specific cases, terms have been arbitrarily selected by the applicant, and in such cases, their meanings will be described in detail in the relevant description of the invention. Therefore, the terms used in this invention should be defined not merely by their names, but based on their meanings and the overall content of the invention. Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application. Hereinafter, embodiments according to the present invention will be described in detail with reference to the attached drawings. FIG. 1 is a flowchart of a picture object recognition method using a deep learning model of the present invention. FIG. 2 is a structural diagram of a deep learning model according to an embodiment of the present invention. FIG. 3 is an image of a PITR subject with a class label and a bounding box for a non-object, which is a stress scale, displayed according to an embodiment of the present invention. First, the present invention includes a recording medium (120) readable by a computer device (100) that records a program for performing a picture object recognition method using a deep learning model. The recording medium (120) may be, for example, a CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM, etc. And the picture object recognition method using a deep learning model of the present invention may be implemented by at least one processor (110) in the computer (100) reading the recording medium (120). Referring to FIG. 1, the deep learning model