Targeting the problems of the insufficient utilization of temporal and spatial information in videos and a lower accuracy rate, this paper proposes a human action recognition method for dynamic videos of emergency rescue based on a spatial-temporal fusion network. A time domain segmentation strategy based on random sampling maintains the overall time domain structure of the video. Considering the spatial-temporal asynchronous relationship, multiple asynchronous motion sequences are increased as input of the temporal convolutional network. spatial-temporal features are fused in convolutional layers to reduce feature loss. Because time series information is crucial for human action recognition, the acquired mid-layer spatial-temporal fusion features are sent into Bidirectional Long Short-Term Memory (Bi-LSTM) to obtain the human movement features in the whole video temporal dimension. Experiment results show the proposed method fully fuses spatial and temporal dimension information and improves the accuracy of human action recognition in dynamic scenes. It is also faster than traditional methods.
The growth of the Internet of Things makes it possible to share information on risky vehicles openly and freely. How to create dynamic knowledge graphs of continually changing risky vehicles has emerged as a crucial technology for identifying risky vehicles, as well as a research hotspot in both artificial intelligence and field knowledge graphs. The node information of the risky vehicle knowledge graph is not rich, and the graph structure plays a major role in its dynamic changes. The paper presents a fusion algorithm based on relational graph convolutional network (R-GCN) and Long Short-Term Memory (LSTM) to build the dynamic knowledge graph of risky vehicles and conducts a comparative experiment on the link prediction task. The results showed that the fusion algorithm based on R-GCN and LSTM had better performance than the other methods such as GCN, DynGEM, ROLAND, and RE-GCN, with the MAP value of 0.2746 and the MRR value of 0.1075. To further verify the proposed algorithm, classification experiments are carried out on the risky vehicle dataset. Accuracy, precision, recall, and F-values were used as heat-tolerance evaluation indexes in classification experiments, the values were 0.667, 0.034, 0.422, and 0.52 respectively.