obtaining a first image frame and a second image frame collected in a movement process of an electronic apparatus;

determining a first feature map corresponding to the first image frame and a second feature map corresponding to the second image frame;

determining a scaled inter-frame geometrical relationship between the first image frame and the second image frame;

determining a reconstruction error between the first feature map and the second feature map based on the inter-frame geometrical relationship; and

determining a depth map corresponding to the first image frame based on the reconstruction error,

wherein the inter-frame geometrical relationship includes a translation distance and a rotation matrix;

wherein the determining a first feature map corresponding to the first image frame and a second feature map corresponding to the second image frame includes respectively performing feature extraction on the first image frame and the second image frame by utilizing a feature extraction branch in a first neural network to obtain the first feature map and the second feature map;

wherein the determining a scaled inter-frame geometrical relationship between the first image frame and the second image frame includes performing pose estimation on the first feature map and the second feature map by utilizing a pose estimation branch in the first neural network to obtain a translation distance and a rotation matrix of image collecting equipment between the first image frame and the second image frame,

wherein the determining a reconstruction error between the first feature map and the second feature map based on the inter-frame geometrical relationship includes calculating the reconstruction error from the second feature map to the first feature map based on the translation distance and the rotation matrix output by the pose estimation branch; and

the method further includes determining a cost volume from the second feature map to the first feature map based on the reconstruction error.