Monocular depth estimation and surface normal vector estimation method based on multi-task network

The invention discloses a monocular depth estimation and surface normal vector estimation method based on a multi-task network. The method comprises the following steps of: collecting multi-scale information by adopting a high-resolution network as a backbone network; outputting features with different resolutions through a high-resolution network, and conducting independent up-sampling on the features respectively to obtain feature maps with the same resolution as the original resolution; connecting the obtained feature maps in series to obtain a multi-scale surface feature, and generating amulti-scale fusion feature; dividing the multi-scale fusion feature into two branch features, and inputting the two branch features into a cross-correlation attention mechanism interaction module to obtain a cross-correlation matrix of learning correlation; inputting the information into a 1*1 continuous convolution layer of each branch feature, then performing softmax operation to obtain two cross-correlation attention g