US 11,854,306 B1
Fitness action recognition model, method of training model, and method of recognizing fitness action
Huapeng Sima, Jiangsu (CN); Hao Jiang, Jiangsu (CN); Hongwei Fan, Jiangsu (CN); Qixun Qu, Jiangsu (CN); Jintai Luan, Jiangsu (CN); and Jiabin Li, Jiangsu (CN)
Assigned to Nanjing Silicon Intelligence Technology Co., Ltd., Nanjing (CN)
Filed by Nanjing Silicon Intelligence Technology Co., Ltd., Jiangsu (CN)
Filed on Jun. 28, 2023, as Appl. No. 18/343,334.
Claims priority of application No. 202211219816.4 (CN), filed on Oct. 8, 2022.
Int. Cl. G06V 40/20 (2022.01); G06V 10/77 (2022.01); G06V 10/764 (2022.01)
CPC G06V 40/23 (2022.01) [G06V 10/764 (2022.01); G06V 10/7715 (2022.01); G06V 2201/12 (2022.01)] 9 Claims
OG exemplary drawing
 
1. A fitness action recognition model, comprising an information extraction layer, a pixel point positioning layer, a feature extraction layer, a vector dimensionality reduction layer, and a feature vector classification layer, wherein
the information extraction layer is configured to obtain image information of a training object in a depth image, the image information comprising a three-dimensional coordinate of human-body key points corresponding to all pixel points in the depth image;
the pixel point positioning layer is configured to perform position estimation on the three-dimensional coordinate of the human-body key points by using a random decision forest, define a body part of the training object as a corresponding body component, and calibrate the three-dimensional coordinate of all human-body key points corresponding to the body component;
the feature extraction layer is configured to extract, based on the three-dimensional coordinate of all the human-body key points, a key-point position feature, a body moving speed feature, and a key-point moving speed feature for action recognition;
the vector dimensionality reduction layer is configured to combine the key-point position feature, the body moving speed feature, and the key-point moving speed feature as a multidimensional feature vector, and perform dimensionality reduction on the multidimensional feature vector; and
the feature vector classification layer is configured to classify the multidimensional feature vector that is performed with dimensionality reduction, to recognize a fitness action of the training object.