US Patent Application 18097080. METHOD FOR GENERATING DEPTH IN IMAGES, ELECTRONIC DEVICE, AND NON-TRANSITORY STORAGE MEDIUM simplified abstract

METHOD FOR GENERATING DEPTH IN IMAGES, ELECTRONIC DEVICE, AND NON-TRANSITORY STORAGE MEDIUM

Organization Name

HON HAI PRECISION INDUSTRY CO., LTD.

Inventor(s)

JUNG-HAO Yang of New Taipei (TW)

CHIH-TE Lu of New Taipei (TW)

CHIN-PIN Kuo of New Taipei (TW)

METHOD FOR GENERATING DEPTH IN IMAGES, ELECTRONIC DEVICE, AND NON-TRANSITORY STORAGE MEDIUM - A simplified explanation of the abstract

This abstract first appeared for US patent application 18097080 titled 'METHOD FOR GENERATING DEPTH IN IMAGES, ELECTRONIC DEVICE, AND NON-TRANSITORY STORAGE MEDIUM

Simplified Explanation

The patent application describes a method and system for generating depth in monocular images using binocular images and instance segmentation labels.

• The method involves acquiring multiple sets of binocular images and building a dataset with instance segmentation labels for content.
• A trained autoencoder network is obtained by training it using the dataset with instance segmentation labels.
• When a monocular image is input into the trained autoencoder network, a first disparity map is obtained.
• The first disparity map is then converted to obtain a depth image corresponding to the monocular image.
• By combining binocular images with instance segmentation images as training data, monocular images can be easily processed to output the disparity map.
• This allows for depth estimation in monocular images by converting the disparity map to a depth image.
• The patent also discloses an electronic device and a non-transitory storage for implementing this method and system.

Original Abstract Submitted

A method and system for generating depth in monocular images acquires multiple sets of binocular images to build a dataset containing instance segmentation labels as to content; training an work using the dataset with instance segmentation labels to obtain a trained autoencoder network; acquiring monocular image, the monocular image is input into the trained autoencoder network to obtain a first disparity map and the first disparity map is converted to obtain depth image corresponding to the monocular image. The method combines binocular images with instance segmentation images as training data for training an autoencoder network, monocular images can simply be input into the autoencoder network to output the disparity map. Depth estimation for monocular images is achieved by converting the disparity map to a depth image corresponding to the monocular image. An electronic device and a non-transitory storage are also disclosed.