Processing

Please wait...

Settings

Settings

Goto Application

1. WO2020064253 - METHODS FOR GENERATING A DEEP NEURAL NET AND FOR LOCALISING AN OBJECT IN AN INPUT IMAGE, DEEP NEURAL NET, COMPUTER PROGRAM PRODUCT, AND COMPUTER-READABLE STORAGE MEDIUM

Publication Number WO/2020/064253
Publication Date 02.04.2020
International Application No. PCT/EP2019/072960
International Filing Date 28.08.2019
IPC
G06K 9/00 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
G06K 9/46 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
36Image preprocessing, i.e. processing the image information without deciding about the identity of the image
46Extraction of features or characteristics of the image
G06K 9/62 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
62Methods or arrangements for recognition using electronic means
CPC
G06K 9/00778
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
00624Recognising scenes, i.e. recognition of a whole field of perception; recognising scene-specific objects
00771Recognising scenes under surveillance, e.g. with Markovian modelling of scene activity
00778Recognition or static of dynamic crowd images, e.g. recognition of crowd congestion
G06K 9/4628
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
36Image preprocessing, i.e. processing the image information without deciding about the identity of the image
46Extraction of features or characteristics of the image
4604Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes, intersections
4609by matching or filtering
4619Biologically-inspired filters, e.g. receptive fields
4623with interaction between the responses of different filters
4628Integrating the filters into a hierarchical structure
G06K 9/6255
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
62Methods or arrangements for recognition using electronic means
6217Design or setup of recognition systems and techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
6255Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries, e.g. user dictionaries
Applicants
  • SIEMENS AKTIENGESELLSCHAFT [DE]/[DE]
Inventors
  • GHOSH, Sanjukta
  • AMON, Peter
  • HUTTER, Andreas
Priority Data
18196304.224.09.2018EP
Publication Language English (EN)
Filing Language English (EN)
Designated States
Title
(EN) METHODS FOR GENERATING A DEEP NEURAL NET AND FOR LOCALISING AN OBJECT IN AN INPUT IMAGE, DEEP NEURAL NET, COMPUTER PROGRAM PRODUCT, AND COMPUTER-READABLE STORAGE MEDIUM
(FR) PROCÉDÉS DE GÉNÉRATION D'UN RÉSEAU NEURONAL PROFOND ET DE LOCALISATION D'UN OBJET DANS UNE IMAGE D'ENTRÉE, RÉSEAU NEURONAL PROFOND, PRODUIT-PROGRAMME INFORMATIQUE ET SUPPORT D'INFORMATIONS LISIBLE PAR ORDINATEUR
Abstract
(EN)
The invention relates to methods (1) for generating a deep neural net (10) and for localising an object (30) in an input image (2), the deep neural net (10), a corresponding computer program product, and a corresponding computer-readable storage medium. The invention proposes to train a discriminative counting model to classify images (2) according to a number of objects (30) of a predetermined type depicted in each of the images (2), and to train a segmentation model to segment images (2) by classifying each pixel according to what image part (30, 31, 35-38, 42, 60) the pixel belongs to. Parts (11) and/or features of both models are combined to form the deep neural net (10), wherein the deep neural net (10) is adapted to generate in a single forward pass a map (14, 16, 52, 58, 63) indicating locations of any objects (30) for each input image (2).
(FR)
L'invention concerne des procédés (1) permettant de générer un réseau neuronal profond (10) et de localiser un objet (30) dans une image d'entrée (2), le réseau neuronal profond (10), un produit-programme informatique correspondant et un support d'informations lisible par ordinateur correspondant. L'invention permet de former un modèle de comptage discriminatif afin de classer des images (2) en fonction d'un nombre d'objets (30) d'un type prédéterminé représenté dans chacune des images (2), et de former un modèle de segmentation afin de segmenter des images (2) par le classement de chaque pixel en fonction de la partie d'image (30, 31, 35-38, 42, 60) à laquelle le pixel appartient. Des parties (11) et/ou des caractéristiques des deux modèles sont combinées afin de former le réseau neuronal profond (10), le réseau neuronal profond (10) étant conçu pour générer, dans un seul passage direct, une carte (14, 16, 52, 58, 63) indiquant des emplacements de n'importe quels objets (30) pour chaque image d'entrée (2).
Also published as
Latest bibliographic data on file with the International Bureau