Method and apparatus for detecting wearing of safety helmet, device and storage medium

Patent metadata

Field	Value
Publication number	US-11610388-B2
Application number	US-202117164613-A
Country	US
Kind code	B2
Filing date	Feb 1, 2021
Priority date	Sep 25, 2020
Publication date	Mar 21, 2023
Grant date	Mar 21, 2023

Abstract

The present application discloses a method and an apparatus for detecting wearing of a safety helmet, a device and a storage medium. The method for detecting wearing of a safety helmet includes: acquiring a first image collected by a camera device, where the first image includes at least one human body image; determining the at least one human body image and at least one head image in the first image; determining a human body image corresponding to each head image in the at least one human body image according to an area where the at least one human body image is located and an area where the at least one head image is located; and processing the human body image corresponding to the at least one head image according to a type of the at least one head image.

First claim

What is claimed is:

1. A method for detecting wearing of a safety helmet, comprising:
acquiring a first image collected by a camera device, wherein the first image comprises at least one human body image;
determining the at least one human body image and at least one head image in the first image;
determining a human body image corresponding to each head image in the at least one human body image according to an area where the at least one human body image is located and an area where the at least one head image is located; and
processing the human body image corresponding to the at least one head image according to a type of the at least one head image;
wherein the determining a human body image corresponding to each head image in the at least one human body image according to an area where the at least one human body image is located and an area where the at least one head image is located comprises:
acquiring matching degree between the area where the at least one head image is located and the area where each human body image is located; and
determining the human body image corresponding to each head image in the at least one human body image according to the matching degree corresponding to each head image;
wherein for a first head image in the at least one head image, determining a human body image corresponding to the first head image in the at least one human body image according to matching degree corresponding to the first head image, comprising:
determining first matching degree in the matching degrees corresponding to the first head image, wherein the first matching degree is a maximum matching degree among the matching degrees corresponding to the first head image; and
when the first matching degree is different from the remaining matching degrees corresponding to the first head image, determining the human body image corresponding to the first matching degree as the human body image corresponding to the first head image;
when the first matching degree is the same as second matching degree, determining the human body image corresponding to the first head image from a first human body image and a second human body image, based on a position of a first area in a second area and a position of the first area in a third area, wherein the first area is an area where the first head image is located, the second area is an area where the first human body image is located, the third area is an area where the second human body image is located; wherein the first human body image is the human body image corresponding to the first matching degree, and the second human body image is the human body image corresponding to the second matching degree.

2. The method according to claim 1 , wherein the determining the at least one human body image and at least one head image in the first image comprises:
processing the first image by a first model to obtain the at least one human body image and the at least one head image, wherein the first model is obtained by learning a plurality of sets of samples, and each set of samples comprises a sample image, a sample human body image and a sample head image.

3. The method according to claim 2 , wherein the processing the first image by a first model to obtain the at least one human body image and the at least one head image comprise:
performing deformable convolution processing on the first image by the first model to obtain image feature information; and
performing processing on the image feature information by the first model to obtain the at least one human body image and the at least one head image.

4. The method according to claim 1 , wherein the method further comprises: processing the human body image corresponding to the first head image according to a type of the first head image, comprising:
adding a label in the first image according to the type of the first head image, wherein the label is used to indicate the human body image corresponding to the first head image;
wherein the type of the first head image is a first type or a second type, the first type is used to indicate that the first head image comprises a safety helmet image, and the second type is used to indicate that the first head image does not comprise the safety helmet image.

5. The method according to claim 4 , wherein the adding a label in the first image according to the type of the first head image comprises:
adding a first label in the first image, when the type of the first head image is the first type; and/or
adding a second label in the first image, when the type of the first head image is the second type.

6. The method according to claim 1 , wherein, after the processing the human body image corresponding to the at least one head image according to a type of the at least one head image, the method further comprises:
sending an alarm instruction to an alarm device, wherein the alarm instruction instructs the alarm device to perform an alarm operation.

7. The method according to claim 1 , wherein the first image is an image collected by the camera device in a preset area.

8. The method according to claim 1 , wherein, after the processing the human body image corresponding to the at least one head image according to a type of the at least one head image, the method further comprises:
displaying the processed image of the human body through a display device.

9. The method according to claim 1 , wherein, the determining the human body image corresponding to the first head image from a first human body image and a second human body image, based on a position of a first area in a second area and a position of the first area in a third area, comprises:
determining the human body image corresponding to the first head image to be the second human body image when the first area is in a middle part of the second area and the first area is in an upper part of the third area.

10. An apparatus for detecting wearing of a safety helmet, comprising: at least one processor and a memory in communicational connection with the at least one processor; wherein,
the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to cause the at least one processor to: acquire a first image collected by a camera device, wherein the first image comprises at least one human body image;
determine at least one human body image and at least one head image in the first image;
determine a human body image corresponding to each head image in the at least one human body image according to an area where the at least one human body image is located and an area where the at least one head image is located; and
process the human body image corresponding to the at least one head image according to a type of the at least one head image;
wherein the instructions further cause the at least one processor to:
acquire matching degree between the area where the at least one head image is located and the area where each human body image is located; and
determine the human body image corresponding to each head image in the at least one human body image according to the matching degree corresponding to each head image;
wherein, for a first head image in the at least one head image, the instructions further cause the at least one processor to:
determine first matching degree in the matching degrees corresponding to the first head image, wherein the first matching degree is a maximum matching degree among the matching degrees corresponding to the first head image; and
when the first matching degree is different from the remaining matching degrees corresponding to the first head image, determine the human body image corresponding to the first matching

Assignees

• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Mao Mingyuan
• Feng Yuan
• Xin Ying
• Yuan Pengcheng
• Zhang Bin
• Lin Shufei
• Wang Xiaodi
• Han Shumin
• Xu Yingbo
• Liu Jingwei
• Wen Shilei
• Zhang Hongwu
• Ding Errui

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title

• G06V20/52

  Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69) · CPC title

• G06N3/045

  Combinations of networks · CPC title