3I Corporation

Farkhod Rustam Ugli Khudayberganov

Ken Kim

Ji Wuck Jung

Mikhail Li

A hole-filling method, performable on a computing device, for providing a three-dimensional virtual model according to a technical aspect of the present application may comprises the operations of: acquiring an original training image and a hole creation training image, wherein the hole creation training image is an image in which at least one hole is created based on the original training image; creating a hole-filling training image by performing hole-filling on the hole creation training image using a neural network; performing spherical transformation on each of the hole-filling training image and the original training images; and training the neural network based on the difference between the spherically transformed hole-filling training image and the spherically transformed original training image.

Pre-processing method for creating 3D virtual model and computing device therefor

A computing device is provided. The computing device includes a memory configured to store one or more instructions; and a processor configured to execute the one or more instructions stored in the memory, in which the processor may execute the one or more instructions to generate a first predicted depth map image for a learning red/green/blue (RGB) image using a neural network, and train the neural network based on a difference between a learning depth map image having depth information and the first predicted depth map image. The learning depth map image may match 1:1 and may be generated based on the same setting change for a basic spherical image.

Depth map image generation method and computing device therefor

A hole filling method for providing a virtual three-dimensional (3D) model, the hole filling method being executable on a computing device and including: acquiring an original learning image and a hole generation learning image for the original learning image, wherein the hole generation learning image is an image in which at least one hole is generated based on the original learning image; generating a hole filling learning image by performing hole filling on the hole generation learning image by using a neural network; performing a spherical transformation on each of the hole filling learning image and the original learning image; and training the neural network based on a difference between the spherically-transformed hole filling learning image and the spherically-transformed original learning image.

Growing community of inventors

Seoul, South Korea

Farkhod Rustam Ugli Khudayberganov