Excrement property estimation model learning device, excrement property estimation device, excrement property estimation model learning method, excrement property estimation method, and program
An excrement property estimation model learning device includes a training data acquisition unit configured to acquire training data including input-side training data including an excrement space which is a predetermined space where there is excrement from an animal being reared, and learning image data that is image data of a captured image obtained by capturing the excrement space, and a correct label indicating whether any abnormal excrement that is abnormal excrement is reflected in an image shown in the learning image data, an excrement property estimation learning model execution unit configured to use an excrement property estimation learning model, which is a machine learning model for estimating a probability that abnormal excrement is reflected in the learning image data, to estimate the probability that abnormal excrement is reflected in the image shown in the learning image data, and an update unit configured to update, on the basis of a difference between an estimation result from the excrement property estimation learning model execution unit and the correct label, the excrement property estimation learning model such that the difference is reduced.
1 . An excrement property estimation model learning device comprising:
a processor; and
a storage medium having computer program instructions stored thereon, wherein the computer program instruction, when executed by the processor perform processing of;
acquiring training data including input-side training data including an excrement space which is a predetermined space where there is excrement from an animal being reared, and learning image data that is image data of a captured image obtained by capturing the excrement space, and a correct label indicating whether any abnormal excrement that is abnormal excrement is reflected in an image shown in the learning image data;
estimating the probability that abnormal excrement is reflected in the image shown in the learning image data by using an excrement property estimation learning model, which is a machine learning model for estimating a probability that abnormal excrement is reflected in the learning image data; and
updating, on the basis of a difference between an estimation result from the excrement property estimation learning model execution unit and the correct label, the excrement property estimation learning model such that the difference is reduced.
2 . The excrement property estimation model learning device according to claim 1 , wherein the learning image data is image data of an extracted excrement image that is an image extracted from a captured image in advance, and is an image of excrement whose horizontal projection area is equal to or greater than a predetermined area.
3 . The excrement property estimation model learning device according to claim 1 , wherein the learning image data is image data of an image after image processing is previously performed on a captured image by a filter that does not allow color of obstacles that are not excrement and whose color and shape are close to those of excrement within a predetermined reference value to pass through.
4 . The excrement property estimation model learning device according to claim 1 , wherein the input-side training data includes information indicating a date and time when the learning image data has been captured.
5 . The excrement property estimation model learning device according to claim 1 , wherein the input-side training data includes information indicating the maximum number of animals whose excrement is present in an excrement space.
6 . The excrement property estimation model learning device according to claim 1 , wherein the input-side training data includes information indicating a feed given to the animal being reared at a predetermined timing prior to a time when an image is captured by an image-capturing device that obtains a captured image.
7 . A non-transitory computer readable medium which stores a program that causes a computer to function as the excrement property estimation model learning device according to claim 1 .
8 . An excrement property estimation model learning method comprising:
acquiring training data including input-side training data including an excrement space which is a predetermined space where there is excrement from an animal being reared, and learning image data that is image data of a captured image obtained by capturing the excrement space, and a correct label indicating whether any abnormal excrement that is abnormal excrement is reflected in an image shown in the learning image data;
estimating the probability that abnormal excrement is reflected in the image shown in the learning image data, by using an excrement property estimation learning model, which is a machine learning model for estimating a probability that abnormal excrement is reflected in the learning image data; and
updating, on the basis of a difference between a result of the estimation in the excrement property estimation learning model execution step and the correct label, the excrement property estimation learning model such that the difference is reduced.