Table 7.

Accuracy comparison of the proposed architecture with various neural network architectures using the LoDVP dataset for detecting abnormal activities.

Video Recognition Architectures	Accuracy [%]
Proposed architecture	93.41
ConvLSTM [21]	92.38
3D Resnet50 [22]	36.19
3D Resnet101 [22]	61.90
3D Resnet152 [22]	90.48