Systems and methods for weakly supervised training of a model for monocular depth estimation

A non-transitory computer-readable medium for semi-supervised training of a depth model for monocular depth estimation and including instructions that when executed by one or more processors cause the one or more processors to:

• train the depth model according to a first stage that is self-supervised and that includes using first training data that comprises pairs of training images, wherein respective ones of the pairs include separate frames depicting a scene from a monocular video, wherein the instructions include instructions to produce first stage loss values that update the depth model and a pose model, wherein the pose model facilitates the first stage according to a structure from motion (SfM) process;
• train the depth model according to a second stage that is weakly supervised and that includes using second training data to produce depth maps according to the depth model, wherein the second training data comprises individual images with corresponding sparse depth data, wherein the instructions to train according to the second stage include instructions to produce second stage loss values that are based, at least in part, on the depth maps and the depth data;
• and
• provide the depth model to infer distances from monocular images in a device.

A method of semi-supervised training of a depth model for monocular depth estimation, comprising:

• training the depth model according to a first stage that is self-supervised and that includes using first training data that comprises pairs of training images, wherein respective ones of the pairs including separate frames depicting a scene of a monocular video, wherein the pairs of training images provide for producing first stage loss values to update the depth model and a pose model, wherein the pose model facilitates the first stage according to a structure from motion (SfM) process;
• training the depth model according to a second stage that is weakly supervised and that includes using second training data to produce depth maps according to the depth model, the second training data comprising individual images with corresponding sparse depth data, the second training data providing for updating the depth model according to second stage loss values that are based, at least in part, on the depth maps and the depth data;
• and
• providing the depth model to infer distances from monocular images in a device.