Multi-view positioning detection method and device

The invention discloses a multi-view positioning detection method and device, and the method comprises the steps: obtaining video data inputted by a plurality of views, and carrying out the extraction through a first neural network, and obtaining a corresponding multi-view first feature map; inputting the multi-view first characteristic spectrum into a second cascade neural network to obtain each level of second characteristic spectrum output by the second cascade neural network, and performing correlation modeling processing on each level of second characteristic spectrum between two adjacent views to obtain similarity matrix data of the two adjacent views; performing aggregation processing on the similarity matrix data of each two adjacent view angles to obtain fused feature data; splicing the multi-view fusion feature data to obtain a global feature map; and determining time sequence nomination and action classification of at least one event according to the global feature map so as to realize positioning