Human body weight recognition method and device based on monocular camera and storage medium

Abstract

The invention relates to artificial intelligence, and provides a human body weight recognition method and device based on a monocular camera and a storage medium, and the method comprises the steps: controlling the monocular camera to focus on a human body and two marking points, enabling the two marking points to be horizontally disposed on background walls at two sides of the human body at intervals, displaying a human body region frame in a photographed image, and enabling the human body to be placed in the human body region frame; judging whether the IOU of the human body minimum boundingrectangle area of the current frame and the human body area frame in the shot picture is greater than a preset intersection threshold, if so, continuing to execute, and otherwise, prompting to focus again; controlling a monocular camera to shoot an image containing a human body and two mark points, inputting the image into the Cascade RCNN network model to identify a human body region, and outputting a mask of a minimum bounding rectangle region of the human body; segmenting a pre-segmented image by taking the human body as the center, sending the pre-segmented image to the trained weight recognition model, and outputting the body length and weight. According to the method, the weight information can be obtained without human body three-dimensional reconstruction, and the method is relatively robust to background interference.

Description

technical field [0001] The present invention relates to artificial intelligence, in particular to a method, device and storage medium for body weight recognition based on a monocular camera. Background technique [0002] With the promotion of 5G and the popularity of smart wearable devices, more and more smart applications appear in people's lives. Especially with the rise of online shopping and various short video software, mobile devices represented by mobile devices provide audio, video, to 3D modeling and other functions. Many smart mobile devices have built-in face recognition and fingerprint recognition. In some occasions, it is necessary to obtain body weight and body size information online. For example, some shopping malls and online shopping platforms have launched virtual fitting apps. During virtual fitting, if you can understand the body size and weight information of the human body, you will be able to recommend suitable clothing for it. Online insurance als...

AI Technical Summary

Problems solved by technology

[0007] Disadvantages: This method can only measure the approximate weight of people with normal body fat percentage, and cannot calculate the exact weight. It is for reference only, and has a great influence on men and women, fat and thin, etc. Since the formula is fixed, it is also suitable for people in different international regions. The accuracy is different, and it is difficult to refine the correction

[0009] Disadvantages: This method requires camera calibration. Due to the need to calculate the relationship between parallax and body height, it has strict requirements on the shooting distance and angle, and is basically not suitable for handheld mobile devices.

[0011] Difficulties: high requirements for algorithms, and poor generalization ability of the model, requiring a large amount of data

[0012] Moreover, none of the above estimation methods can adapt to the measurement of the human body on the mobile terminal, which is neither accurate nor convenient enough. Up to now, there is no general human body weight recognition system on the mobile terminal

Images