Anomaly Event Detection in Security Surveillance Using Two-Stream Based Model

Figure 10

The visualization results of our method on testing videos on ShanghaiTech. The first row exhibits the visualization results obtained by our best model variant, and the purple rectangles represent the ground truth time period in which the anomaly event occurred. The second row shows the visualization results of different variants of our model, including results of ResNet50, ResNet50, ResNet150, and the best model variants, respectively, and Flow modalities in the video. Moreover, our proposed solution can provide inspiration for other video-related tasks, including video classification, video segmentation, video tracking and video detection, through bistream setting to obtain the improved.