Extension usage examples:
Here's how our browser extension sees the article:
多尺度方法结合卷积神经网络的行为识别
Source: cea.ceaj.org
Article summary:
1. A recognition model based on convolutional neural networks with multi-scale convolution operators is proposed to calculate the temporal and spatial features of behavior sequence samples.
2. The convolution operator in each layer of the convolutional neural network is extended to a multi-scale convolution operator to explore the role of skeletal points with different adjacency relationships in describing human behavior.
3. Experiments on MSR-Action3D dataset and HDM05 dataset achieved better recognition rate.
Article analysis:
The article provides a detailed description of a recognition model based on convolutional neural networks with multi-scale convolution operators for calculating temporal and spatial features of behavior sequence samples. The article does not provide any evidence or data to support its claims, which makes it difficult to assess the trustworthiness and reliability of the article. Additionally, there is no discussion about potential risks associated with this model, such as privacy concerns or potential bias in the data used for training the model. Furthermore, there is no mention of any counterarguments or alternative approaches that could be used for recognizing behaviors from sequence samples. Finally, there is no indication that both sides of an argument have been presented equally, which could lead to partiality in reporting.