IP Library Granted Patent US 12711376
Granted Patent B2
US 12711376 · App. 18/052,248 · Granted Aug 18, 2026

Pseudo-data generation apparatus, pseudo-data generation method, learning apparatus and learning method

Inventor: Hidenori Takeshima (Tokyo, JP)
Assignee: Canon Kabushiki Kaisha
G06N3/08G06T2207/10072
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12711376
App. No.
18/052,248
Granted
Aug 18, 2026
Kind
B2
Abstract

According to one embodiment, a pseudo-data generation apparatus includes processing circuitry. The processing circuitry acquires one or more pieces of partial observation data that form part of whole observation data. The processing circuitry generates pseudo-whole observation data by inputting the one or more pieces of partial observation data to a function, the pseudo-whole observation data being pseudo-data of the whole observation data. The function is optimized by training so that partial observation data for training and pseudo-partial observation data for training resemble each other, the pseudo-partial observation data for training being obtained by converting the pseudo-whole observation data for training.

Claims (15)

1 . A pseudo-data generation apparatus for generating whole observation data which is unavailable and impossible to observe, the pseudo-data generation method comprising:

processing circuitry configured to:

acquire one or more pieces of partial observation data that form part of the whole observation data, the whole observation data being three-dimensional volume magnetic resonance moving image data; and

generate pseudo-whole observation data by inputting the one or more pieces of partial observation data to a function, the pseudo-whole observation data being pseudo-data of the whole observation data,

wherein the function is optimized by training, without using the unavailable whole observation data, so that partial observation data for training and pseudo-partial observation data for training resemble each other, the pseudo-partial observation data for training being obtained by converting pseudo-whole observation data for training, and

wherein the pseudo-whole observation data for training is generated using the function, from the partial observation data for training.

2 . The pseudo-data generation apparatus according to claim 1 , wherein the function is a generator trained using a conditional generative adversarial network, a decoder trained using a conditional variational auto encoder, or a model trained using a conditional diffusion model.

3 . The pseudo-data generation apparatus according to claim 1 , wherein the processing circuitry is further configured to:

acquire, as multiple pieces of the partial observation data, a plurality of images captured while shifting a focus, and

generate three-dimensional volume data as the pseudo-whole observation data from the multiple pieces of the partial observation data, the three-dimensional volume data including depth information.

4 . A pseudo-data generation method for generating whole observation data which is impossible to observe and unavailable, the pseudo-data generation method comprising:

acquiring one or more pieces of partial observation data that form part of whole observation data, the whole observation data being three-dimensional volume magnetic resonance moving image data; and

generating pseudo-whole observation data by inputting the one or more pieces of partial observation data to a function, the pseudo-whole observation data being pseudo-data of the whole observation data,

wherein the function is optimized by training, without using the unavailable whole observation data, so that partial observation data for training and pseudo-partial observation data for training resemble each other, the pseudo-partial observation data for training being obtained by converting pseudo-whole observation data for training, and

wherein the pseudo-whole observation data for training is generated using the function, from the partial observation data for training.