WO2023072744 - METHODS AND SYSTEMS FOR GENERATING SEGMENTATION MASKS
National phase entry:
Publication Number
WO/2023/072744
Publication Date
04.05.2023
International Application No.
PCT/EP2022/079332
International Filing Date
21.10.2022
Title **
[English]
METHODS AND SYSTEMS FOR GENERATING SEGMENTATION MASKS
[French]
PROCÉDÉS ET SYSTÈMES DE GÉNÉRATION DE MASQUES DE SEGMENTATION
Applicants **
SIEMENS AKTIENGESELLSCHAFT
Inventors
JANISCH, Lucas
LUTZ, Benjamin Samuel
Priority Data
21205589.1
29.10.2021
EP
Application details
| Total Number of Claims/PCT | * |
| Number of Independent Claims | * |
| Number of Priorities | * |
| Number of Multi-Dependent Claims | * |
| Number of Drawings | * |
| Pages for Publication | * |
| Number of Pages with Drawings | * |
| Pages of Specification | * |
| * | |
| Number of Office Actions | * |
| * | |
International Searching Authority |
EPO
* |
| Recordal of a Change of the Applicant's Name/Address |
Change of Applicant's Name and Address
* |
| Type of Assignment |
The Standard Agent's Assignment
* |
| Applicant's Legal Status |
Legal Entity
* |
| * | |
| * | |
| * | |
| * | |
| * | |
| Entry into National Phase under |
Chapter II
* |
| * | |
| * | |
| * | |
| Patent Delivery |
Send the Letters Patent by Courier
* |
| Translation |
|
* The data is based on automatic recognition. Please verify and amend if necessary.
** IP-Coster compiles data from publicly available sources. If this data includes your personal information, you can contact us to request its removal.
Quotation for National Phase entry
| Country | Stages | Total | |
|---|---|---|---|
| China | Filing, Examination, Granting | 2256 | |
| EPO | Filing, Examination, Granting | 9795 | |
| Japan | Filing, Examination, Granting | 2232 | |
| South Korea | Filing, Examination, Granting | 2185 | |
| USA | Filing, Examination, Granting | 5940 |
Total:
22,408
The term for entry into the National Phase has expired. This quotation is for informational purposes only
Contact Us
Abstract[English]
Method for generating a segmentation mask (MIm, CMIm1, CMIm2) of at least one image (IM), the method comprising A) Generating a plurality of superpixels (SP) for the at least one image (IM), B) Automatically generating labels for the generated plurality of the superpixels (SP), wherein said automatically generating labels comprises generating said labels for training a semantic segmentation model based on a plurality of segmentation masks (MIm, CMIm1) by: generating a label for each superpixel (SP) by identifying the most similar reference superpixel (RSP11,..., RSP34) from a reference data set of reference superpixels, wherein each reference superpixel (RSP11,..., RSP34) is associated with a class (Class 1, Class 2, Class 3,... Class n) for supervised training of said semantic segmentation model, wherein the labeled superpixels form the segmentation mask (MIm, CMIm1) of the at least one image (Im); C) Computer-aided checking of the generated segmentation mask (MIm) for correctness, wherein said checking for correctness comprises labeling unlabeled superpixels and correcting labels of incorrectly labeled superpixels by assigning the incorrectly labeled superpixels to the correct class.[French]
Procédé de génération d'un masque de segmentation (MIm, CMIm1, CMIm2) d'au moins une image (IM), le procédé consistant A) à générer une pluralité de superpixels (SP) correspondant à la ou aux images (IM), B) à générer automatiquement des étiquettes correspondant à la pluralité de superpixels générée (SP), ladite génération automatique d'étiquettes consistant à générer lesdites étiquettes pour former un modèle de segmentation sémantique sur la base d'une pluralité de masques de segmentation (MIm, CMIm1) : par la génération d'une étiquette pour chaque superpixel (SP) en identifiant le superpixel de référence le plus similaire (RSP11..., RSP34) à partir d'un ensemble de données de référence de superpixels de référence, chaque superpixel de référence (RSP11..., RSP34) étant associé à une classe (classe 1, classe 2, classe 3, ..., classe n) pour la formation supervisée dudit modèle de segmentation sémantique, les superpixels étiquetés formant le masque de segmentation (MIm, CMIm1) de ladite image (Im) ; C) à effectuer la vérification assistée par ordinateur de l'exactitude du masque de segmentation généré (MIm), ladite vérification de l'exactitude comprenant l'étiquetage de superpixels non étiquetés et la correction d'étiquettes de superpixels étiquetés de manière incorrecte par l'attribution des superpixels incorrectement étiquetés à la classe correcte.