Learning device, parameter adjustment method and recording medium
A learning device includes a learning model for a still image. The learning model includes a mask generation means configured to generate a first object mask identifying an area in which an object exists in a still image, for each individual object. A first parameter including at least one parameter used for processing of generating the first object mask is adjusted based on a first loss. The first loss indicates a difference of the first object mask with respect to a second object mask identifying an area in which an object exists in a moving image including the still image, for each individual object.
1 . A learning device comprising:
a memory configured to store instructions; and
a processor functioning as a learning model for a still image, the processor configured to execute the instructions to:
generate a first object mask identifying an area in which an object exists in the still image, for each individual object,
wherein a first parameter including at least one parameter used for processing of generating the first object mask is adjusted based on a first loss, the first loss indicating a difference of the first object mask with respect to a second object mask identifying an area in which an object exists in a moving image including the still image, for each individual object,
wherein the processor is further configured to execute the instructions to extract a first feature quantity representing a feature of the entire still image,
wherein a second parameter including at least one parameter used for processing of extracting the first feature quantity is adjusted based on a second loss, the second loss indicating a difference of the first feature quantity with respect to a second feature quantity representing a feature of the entire moving image.
2 . The learning device according to claim 1 , wherein the first parameter is adjusted in a manner that the first loss becomes 0.
3 . The learning device according to claim 1 , wherein the first parameter and the second parameter are adjusted in a manner that both the first loss and the second loss become 0.
4 . The learning device according to claim 1 ,
wherein the processor is further configured to execute the instructions to acquire a first object representation for each individual object in the still image,
wherein a third parameter including at least one parameter used for processing of acquiring the first object representation is adjusted based on a third loss, the third loss indicating a difference of the first object representation with respect to a second object representation, for each individual object in the moving image.
5 . The learning device according to claim 4 , wherein the first parameter and the third parameter are adjusted in a manner that both the first loss and the third loss become 0.
6 . The learning device according to claim 1 ,
wherein the processor is configured to execute the instructions to:
extract the first feature quantity;
acquire a first object representation for each individual object in the still image based on the first feature quantity; and
generate the first object mask based on the first object representation,
wherein a third parameter including at least one parameter used for processing of acquiring the first object representation is adjusted based on a third loss, the third loss indicating a difference of the first object representation with respect to a second object representation, for each individual object in the moving image.
7 . The learning device according to claim 6 , wherein the first parameter, the second parameter and the third parameter are adjusted in a manner that all the first loss, the second loss and the third loss become 0.
8 . A parameter adjustment method applied to a learning model for a still image that generates a first object mask identifying an area where an object exists in the still image for each individual object, the method comprising:
adjusting a first parameter including at least one parameter used for processing of generating the first object mask based on a first loss, the first loss indicating a difference of the first object mask with respect to a second object mask identifying an area where an object exists in a moving image including the still image, for each individual object,
wherein further executing the instructions to extract a first feature quantity representing a feature of the entire still image,
wherein a second parameter including at least one parameter used for processing of extracting the first feature quantity is adjusted based on a second loss, the second loss indicating a difference of the first feature quantity with respect to a second feature quantity representing a feature of the entire moving image.
9 . A non-transitory computer-readable recording medium storing a program, the program causing a computer to execute processing for a learning model for a still image that generates a first object mask identifying an area where an object exists in the still image for each individual object, the processing comprising:
adjusting a first parameter including at least one parameter used for processing of generating the first object mask based on a first loss, the first loss indicating a difference of the first object mask with respect to a second object mask identifying an area where an object exists in a moving image including the still image, for each individual object,
wherein further executing the instructions to extract a first feature quantity representing a feature of the entire still image,
wherein a second parameter including at least one parameter used for processing of extracting the first feature quantity is adjusted based on a second loss, the second loss indicating a difference of the first feature quantity with respect to a second feature quantity representing a feature of the entire moving image.