Table 1.
Comparison of highway abnormal event detection models.
Fig 1.
Technical process of this article.
Fig 2.
Heterogeneous graph neural network structure.
Table 2.
Experimental model parameter settings.
Fig 3.
Monitoring of image and video feature information under traffic congestion in tunnel.
Fig 4.
Fused feature data.
Fig 5.
Typical prediction analysis scenario.
Table 3.
Prediction results of traffic abnormal events under different fusion information inputs.
Fig 6.
Pulse signal diagram of dynamic abnormal event detection results in four different scenarios.
Table 4.
Overall performance indicators of the algorithm model in this article for dynamic detection of traffic anomalies on highways.
Fig 7.
Changes in various performance indicators in the time dimension under normal scenarios.
Fig 8.
Changes in various performance indicators in the time dimension under accident scenarios.
Fig 9.
Changes in various performance indicators in the time dimension under congested scenarios.
Fig 10.
Changes in various performance indicators in the time dimension under the influence of severe weather conditions.
Table 5.
Comparison of static detection performance (Traffic flow scenario).
Table 6.
Comparison of static prediction performance (Traffic flow prediction scenario).
Fig 11.
Comparison of detection and prediction performance of different algorithm models in static scenes.
Table 7.
Comparison of dynamic detection performance (Accident detection scenario).
Table 8.
Comparison of dynamic prediction performance (Accident prediction scenarios).
Fig 12.
Performance Comparison of Dynamic Detection and Prediction in Different Scenarios.
Table 9.
Cross-dataset generalization performance.
Table 10.
Model robustness testing results.
Table 11.
Results of ablation experiments.