Electronic device assigning recognition model to computing unit and operating method thereof
An electronic device recognizing an object is provided. The electronic device includes a plurality of computing units, a memory, and a processor configured to control at least one of the plurality of computing units such that object information about objects obtained by recognizing the objects existing in a space by using a first recognition model, divide the space into a plurality of subset spaces, based on the object information, determine at least one recognition model, based on characteristic information of each of the subset spaces, assign the determined recognition model to one computing unit, based on characteristic information of each of a plurality of computing units and characteristic information of the determined recognition model, and control the plurality of computing units to perform object recognition by using the determined recognition model and the one computing unit in each of the subset spaces.
1. An electronic device comprising:
a plurality of computing processors;
memory storing a plurality of recognition models and one or more instructions; and
a processor communicatively coupled to the plurality of computing processors and the memory,
wherein the one or more instructions, when executed by the processor, cause the electronic device to:
control at least one of the plurality of computing processors to obtain object information about one or more objects by recognizing the one or more objects existing in a space of a space map by using a first recognition model from among the plurality of recognition models, the space map corresponding to a space map of a structure,
based on the object information, divide the space of the space map into a plurality of subset spaces,
based on characteristic information of the plurality of subset spaces, determine at least one recognition model to be used in each of the plurality of subset spaces from among the plurality of recognition models, respectively,
assign the determined at least one recognition model to one of the plurality of computing processors based on an operation speed of each of the plurality of computing processors and at least one of a number of classes that are recognizable, an operation amount, or a runtime memory of the determined at least one recognition model, and
control the plurality of computing processors to perform object recognition by using the determined at least one recognition model and the one of the plurality of computing processors, to which the determined at least one recognition model is assigned, in response to the electronic device moving through each of the plurality of subset spaces, respectively.
2. The electronic device of claim 1 , further comprising:
a sensor configured to obtain spatial structure information of the space by scanning the space,
wherein the one or more instructions, when executed by the processor, further cause the processor to:
generate the space map based on the spatial structure information, and
based on the space map and the object information, divide the space into the plurality of subset spaces.
3. The electronic device of claim 2 , wherein the one or more instructions, when executed by the processor, further cause the processor to:
based on whether the structure of the space changes, determine whether to perform a scanning of the space, and
based on determining to perform the scanning of the space, control the sensor to scan the space while moving through the structure of the space.
4. The electronic device of claim 1 , wherein classes recognizable by the first recognition model are higher classes than classes recognizable by the determined at least one recognition model with respect to each of the plurality of subset spaces.
5. The electronic device of claim 1 , wherein types of objects that are recognizable by the determined at least one recognition model with respect to each of the plurality of subset spaces are less than types of objects that are recognizable by the first recognition model.
6. The electronic device of claim 1 , wherein the one or more instructions, when executed by the processor, further cause the processor to:
in response to the operation amount of the determined at least one recognition model being equal to or greater than a first operation value, assign the determined at least one recognition model to a first determined computing processor, from among the plurality of computing processors, having the operation speed equal to or greater than a second operation value, and
in response to the operation amount of the determined at least one recognition model being less than the first operation value, assign the determined at least one recognition model to a second determined computing processor, from among the plurality of computing processors, having the operation speed less than the second operation value.
7. The electronic device of claim 1 , wherein the one or more instructions, when executed by the processor, further cause the processor to:
divide the space into a first subset space classified as a first group and a second subset space classified as a second group, based on the object information,
based on characteristics of the first subset space, determine a second recognition model, from among recognition models corresponding to the first group, to be used in the first subset space, and
based on characteristics of the second subset space, determine a third recognition model, from among recognition models corresponding to the second group, to be used in the second subset space.
8. The electronic device of claim 7 , wherein the one or more instructions, when executed by the processor, further cause the processor to:
assign the second recognition model to a first computing processor from among the plurality of computing processors and assign the third recognition model to a second computing processor, based on characteristic information of the plurality of computing processors, characteristic information of the second recognition model, and characteristic information of the third recognition model,
control to perform object recognition in the first subset space by using the first computing processor and the second recognition model, and
control to perform object recognition in the second subset space by using the second computing processor and the third recognition model.
9. The electronic device of claim 7 , further comprising:
a communicator,
wherein the one or more instructions, when executed by the processor, further cause the processor to:
control the communicator to transmit, to an external device, a request for recognition models corresponding to characteristic information of the plurality of computing processors, the first group, and the second group,
receive, from the external device, the recognition models corresponding to the first group and the second group, and
store the recognition models in the memory, and
wherein the recognition models corresponding to the first group and the second group received from the external device are determined based on the characteristic information of the plurality of computing processors.
10. An operating method performed by an electronic device, the operating method comprising:
obtaining, by the electronic device, object information about one or more objects by recognizing the one or more objects existing in a space of a space map by using a first recognition model from among a plurality of recognition models, the space map corresponding to a space map of a structure;
based on the object information, dividing, by the electronic device, the space of the space map into a plurality of subset spaces;
based on characteristic information of the plurality of subset spaces, determining, by the electronic device, at least one recognition model to be used in each of the plurality of subset spaces, from among the plurality of recognition models, respectively;
assigning, by the electronic device, the determined at least one recognition model to one of a plurality of computing processors, based on an operation speed of each of each of the plurality of computing processors and at least one of a number of classes that are recognizable, an operation amount, or a runtime memory of the determined at least one recognition model; and
performing, by the electronic device, object recognition by using the determined at least one recognition model and the one of the plurality of computing processors to which the determined at least one recognition model is assigned, in response to the electronic device moving through each of the plurality of subset spaces, respectively.
11. The operating method of claim 10 , further comprising:
obtaining, by the electronic device, spatial structure information about the space by scanning the space; and
generating, by the electronic device, the space map based on the spatial structure information,
wherein the dividing of the space of the space map into the plurality of subset spaces comprises dividing the space into the plurality of subset spaces based on the space map and the object information.
12. The operating method of claim 11 , further comprising:
based on whether the structure of the space changes, determining, by the electronic device, whether to perform a scanning of the space,
wherein the obtaining of the spatial structure information about the space comprises scanning the space in response to determining to perform the scanning of the space.
13. The operating method of claim 10 , wherein classes recognizable by the first recognition model are higher classes than classes recognizable by the determined at least one recognition model with respect to each of the plurality of subset spaces.
14. A non-transitory computer-readable recording medium having stored thereon a program for performing the operating method of claim 10 .
15. The electronic device of claim 1 , wherein the one or more instructions, when executed by the processor, further cause the processor to:
based on characteristic information of the plurality of computing processors and characteristic information of the determined at least one recognition model, load at least one second recognition model among the plurality of recognition models onto a first computing processor among the plurality of computing processors while the electronic device moves through a first subset space among the plurality of subset spaces to perform object recognition, and
based on the characteristic information of the plurality of computing processors and the characteristic information of the determined at least one recognition model, load at least one third recognition model among the plurality of recognition models onto a second computing processor among the plurality of computing processors while the electronic device moves through a second subset space among the plurality of subset spaces to perform object recognition.