Location determination apparatus, location determination method and computer program

1. A position determination apparatus for determining, from a first image that is an image including at least a desired first subject, a position of the desired first subject, the position determination apparatus comprising:

a processor;

a storage medium having computer program instructions stored thereon, when executed by the processor, perform to:

acquire a first feature amount representing a predetermined feature of the first image; and

determines the position of the desired first subject using an estimated first feature map, where an estimation unit in which a second feature amount representing the predetermined feature obtained from a second image including a second subject is associated with a selected feature map that is a feature map corresponding to the second subject among a plurality of feature maps that are a set of feature maps that are maps representing features of the second image; and the estimated first feature map is estimated by the estimation unit to be associated with the first feature amount.

2. The position determination apparatus according to claim 1 , wherein

a feature map included in the plurality of feature maps representing the features of the second image is an output obtained by performing a first convolution process on the second image, or an output obtained by further performing a second convolution process on the output obtained by performing the convolution process.

3. The position determination apparatus according to claim 2 , wherein:

the first convolution process and the second convolution process are performed in intermediate layers of a learned neural network; and

the feature map included in the plurality of feature maps is an output value of the intermediate layers.

4. The position determination apparatus according to claim 3 , wherein the selected feature map is a feature map that reacts more strongly to an area including the second subject than an area other than the second subject in the second image.

5. A position determination method for determining, from a first image that is an image including at least a desired first subject, a position of the desired first subject, the position determination method comprising:

a feature amount acquisition step of acquiring a first feature amount representing a predetermined feature of the first image;

an estimation step in which a second feature amount representing the predetermined feature obtained from a second image including a second subject is associated with a selected feature map that is a feature map corresponding to the second subject among a plurality of feature maps that are a set of maps representing features of the second image; and

a position determination step of determining the position of the desired first subject from the first image using an estimated first feature map that is estimated in the estimation step to be associated with the first feature amount.

6. A non-transitory computer readable medium including instructions executable by one or more processors to:

acquiring a first feature amount representing a predetermined feature of the first image; and determine the position of the desired first subject from the first image using an estimated first feature map that is estimated in an estimation step to be associated with the first feature amount, where the estimation step in which a second feature amount representing the predetermined feature obtained from a second image including a second subject is associated with a selected feature map that is a feature map corresponding to the second subject among a plurality of feature maps that are a set of maps representing features of the second image; and

a position determination step of.