Codierung von Trainingsdaten zum Trainieren eines Neuronalen Netzes

Details

Veröffentlichungs-Nr.

EP4266260

Aktenzeichen

EP22169005

Anmeldetag

20. April 2022

Veröffentlichung

7. Januar 2026

Erteilung

7. Januar 2026

Rechtsraum

EP

IPC

G06V10/774G06V10/82

Abstract

The present invention generally relates to a method for encoding training data for training of a neural network, the method comprising: obtaining (S102) training data including multiple datasets (402, 404) where each dataset comprises images annotated with at least one respective object class and where a number of background classes is less than the number of datasets, forming (S104), for each of the multiple datasets (402, 404), an individual background class (B1, B2) associated with the at least one respective object class annotated in each of the multiple datasets (402, 404); encoding (S106) the images of the multiple datasets to be associated with their respective individual background class; encoding (S107) image patches belonging to annotated object classes in the images of the multiple datasets to be associated with their respective object class; encoding (S108) each of the multiple datasets, to include an ignore attribute ("ignore") to object classes (104) that are annotated only in the other datasets (2) of the multiple datasets and to background classes (108) formed for the other datasets of the multiple datasets, the ignore attribute indicating that the assigned object class and background classes do not contribute in adapting the neural network in training using the respective dataset; and providing (S110) the encoded training data for training of a neural network.

Anmelder

Firma

Axis AB

Land

🇸🇪 Schweden

Fachgebiete