Determining parameters of anchor boxes used in a sliding window method when a bounding box and a class of an object in an image are detected using a neural network and the sliding window method

Patent metadata

Field	Value
Publication number	US-12299950-B2
Application number	US-201917778926-A
Country	US
Kind code	B2
Filing date	Dec 6, 2019
Priority date	Dec 6, 2019
Publication date	May 13, 2025
Grant date	May 13, 2025

Abstract

A detection object analysis unit ( 4 ) is a parameter determination apparatus that determines parameters of a plurality of anchor boxes to be used in a sliding window method when a bounding box and a class of an object in an image are detected using a neural network and the sliding window method. The detection object analysis unit ( 4 ) includes a distribution generation unit ( 11 ) that generates distribution information of parameters of bounding boxes indicated by object specifying information of a plurality of pieces of learning data. The detection object analysis unit ( 4 ) includes a clustering processing unit ( 12 ) that generates a plurality of clusters by clustering the distribution information. The detection object analysis unit ( 4 ) includes a parameter determination unit ( 13 ) that determines the parameters of the plurality of anchor boxes based on the plurality of clusters.

First claim

What is claimed is:

1. A parameter determination apparatus for determining parameters of a plurality of anchor boxes to be used in a sliding window method when a bounding box and a class of an object in an image are detected using a neural network and the sliding window method, wherein
the neural network outputs a bounding box and a class of an object in an input image using a learned model learned using a learning data set formed so as to include a plurality of pieces of learning data, each piece of learning data of the plurality of pieces of learning data including an image that shows at least one object and object specifying information indicating a bounding box in the image that shows the at least one object and a class of the at least one object, the parameter determination apparatus comprising:
at least one memory storing computer-executable instructions; and
at least one processor configured to access the at least one memory and execute the computer-executable instructions to:
generate distribution information of parameters of the bounding boxes indicated by the object specifying information of the plurality of pieces of learning data;
generate a plurality of clusters by clustering the distribution information, wherein generating the plurality of clusters includes clustering the distribution information in such a way that within-cluster variance of each cluster of the plurality of clusters increases in proportion to a scale of the bounding boxes indicated by the object specifying information of the plurality of pieces of learning data; and
determine the parameters of the plurality of anchor boxes based on the plurality of clusters.

2. The parameter determination apparatus according to claim 1 , wherein
generating the distribution information includes generating distribution information on vertical dimensions and horizontal dimensions of the parameters of the bounding boxes indicated by the object specifying information of the plurality of pieces of learning data, and
the at least one processor determines vertical dimensions and horizontal dimensions of the plurality of anchor boxes.

3. A computer-implemented parameter determination method for determining parameters of a plurality of anchor boxes to be used in a sliding window method when a bounding box and a class of an object in an image are detected using a neural network and the sliding window method, wherein
the neural network outputs a bounding box and a class of an object in an input image using a learned model learned using a learning data set formed so as to include a plurality of pieces of learning data, each piece of learning data of the plurality of pieces of learning data including an image that shows at least one object and object specifying information indicating a bounding box in the image that shows the at least one object and a class of the at least one object, the parameter determination method being performed by at least one processor executing stored instructions to perform steps comprising:
generating distribution information of parameters of the bounding boxes indicated by the object specifying information of the plurality of pieces of learning data;
generating a plurality of clusters by clustering the distribution information, wherein generating the plurality of clusters includes clustering the distribution information in such a way that within-cluster variance of each cluster of the plurality of clusters increases in proportion to a scale of the bounding boxes indicated by the object specifying information of the plurality of pieces of learning data; and
determining the parameters of the plurality of anchor boxes based on the plurality of clusters.

4. A non-transitory computer readable medium storing a program executable by a computer to perform the parameter determination method according to claim 3 .

Assignees

• Nec Corp

Inventors

• Kamimura Junichi

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06V10/72

  Data preparation, e.g. statistical preprocessing of image or video features · CPC title

• G06V10/763

  Non-hierarchical techniques, e.g. based on statistics of modelling distributions · CPC title