HSFV-based Action Recognition Using Recurrent Neural Networks
Citations Over Time
Abstract
Human action recognition is one of the basic problems in the field of computer vision, which has a wide range of applications in video surveillance, sports analysis, medical care. Since the skeleton data can not be easily affected by background and views and has small computational cost, skeleton-based action recognition has attracted a lot of researchers’ interest. In recent years, some researchers have proposed to use the joints connection methods to mine the intrinsic correlation between skeleton joints as the feature representation of motion skeleton, and good results are achieved. However, there is still the problem of confusion when distinguishing actions with similar motion segments. To solve this problem, a human skeleton feature vector model (HSFV) is proposed in this paper. By constructing the feature extraction reference frame, the model uses OKS indicator to calculate and generate the feature vector describing the human posture. The feature vectors are input into the recurrent neural network, the experimental results on public and self-built datasets show that the human skeleton feature vector model based action recognition method proposed in this paper can distinguish actions with similar motion segments, and has the advantages of simple training and small calculation cost. It has broad application prospects in the fields of process detection, motion evaluation and so on.
Related Papers
- → Genetical studies on the skeleton of the mouse(1953)81 cited
- → Rate of Renewal of the Fish Skeleton(1945)8 cited
- → The Skeleton(1978)1 cited
- [Skeletal characteristics of the hand in cervical cancer].(1978)
- Absorbed fractions and dose factors for a model of the mouse skeleton(2016)