Method for obtaining depth images, electronic device, and storage medium

An electronic device comprising:

• at least one processor;
• and
• a storage device coupled to the at least one processor and storing instructions for execution by the at least one processor to cause the at least one processor to:
• obtain a first image and a second image, the first image and the second image are adjacent frame images;
• obtain a predicted depth map of the first image based on a deep learning network model, and calculate a first error value of the predicted depth map by using a first preset loss function;
• determine a first transformation matrix between the first image and the second image;
• obtain an instance segmentation image by performing an instance segmentation process on the first image, and obtain a first mask image and a second mask image by performing a mask process on the instance segmentation image;
• obtain a target transformation matrix by performing an averaging process on the first transformation matrix according to the first mask image and the second mask image;
• convert the predicted depth map into a first point cloud image based on internal parameters of a camera device, convert the first point cloud image into a second point cloud image based on the target transformation matrix, and convert the second point cloud image into a third image;
• calculate a second error value between the second image and the third image by using a second preset loss function;
• obtain a target deep learning network model by adjusting the deep learning network model according to the first error value and the second error value;
• and
• input at least one image to be detected into the target deep learning network model, and obtain a depth image corresponding to each of the at least one image to be detected.

A computer-readable storage medium having instructions stored thereon, when the instructions are executed by a processor of an electronic device, the processor is configured to perform a method for obtaining depth images, wherein the method comprises:

• obtaining a first image and a second image, the first image and the second image being adjacent frame images;
• obtaining a predicted depth map of the first image based on a deep learning network model, and calculating a first error value of the predicted depth map by using a first preset loss function;
• determining a first transformation matrix between the first image and the second image;
• obtaining an instance segmentation image by performing an instance segmentation process on the first image, and obtaining a first mask image and a second mask image by performing a mask process on the instance segmentation image;
• obtaining a target transformation matrix by performing an averaging process on the first transformation matrix according to the first mask image and the second mask image;
• converting the predicted depth map into a first point cloud image based on internal parameters of a camera device, converting the first point cloud image into a second point cloud image based on the target transformation matrix, and converting the second point cloud image into a third image;
• calculating a second error value between the second image and the third image by using a second preset loss function;
• obtaining a target deep learning network model by adjusting the deep learning network model according to the first error value and the second error value;
• and
• inputting at least one image to be detected into the target deep learning network model, and obtaining a depth image corresponding to each of the at least one image to be detected.