Learning device and recognition device for first sensor and second sensor

1 . An information processing apparatus, comprising:

a central processing unit (CPU) is configured to:

perform a first recognition process based on a first signal from a first sensor of a plurality of sensors;

perform a second recognition process based on a second signal from a second sensor of the plurality of sensors, wherein

a first reading unit of the first sensor is different from a second reading unit of the second sensor; and

generate control information to control the second recognition process, wherein

the control information is generated based on one of:

a dataset used to train in the first recognition process, or

output of the first recognition process.

2 . The information processing apparatus according to claim 1 , wherein

the first reading unit of the first sensor is one frame, and

the second reading unit of the second sensor is smaller than the one frame.

3 . The information processing apparatus according to claim 1 , wherein the CPU is further configured to:

estimate, based on each of the dataset and a control range indicating a range to control the first sensor, a statistic corresponding to the control range of the first signal; and

generate the control information based on the estimated statistic.

4 . The information processing apparatus according to claim 3 , wherein the CPU is further configured to

generate, based on the statistic, the control information to control a timing of a reading operation from the second sensor.

5 . The information processing apparatus according to claim 4 , wherein

the statistic indicates an appearance frequency for each line of a plurality of lines of a recognition target included in the first signal, and

the CPU is further configured to generate the control information to control the timing based on the appearance frequency.

6 . The information processing apparatus according to claim 3 , wherein

the CPU is further configured to generate the control information based on a control constraint, and

the control constraint is a constraint to control the first sensor.

7 . The information processing apparatus according to claim 3 , wherein the CPU is further configured to:

sample information at a sampling position, based on the control information, wherein

the information is obtained by addition of time series information to the dataset;

update the control information based on the sampled information; and

update the time series information and the sampling position, based on the updated control information.

8 . The information processing apparatus according to claim 1 , wherein

the CPU is further configured to control a reading operation of the second signal from the second sensor based on distillation, and

the distillation is based on an output of the first recognition process.

9 . The information processing apparatus according to claim 1 , wherein

the CPU is further configured to train the second recognition process based on distillation, and

the distillation is based on the output of the first recognition process.

10 . An information processing apparatus, comprising:

a central processing unit (CPU) configured to:

perform a first recognition process based on a first signal from a first sensor of a plurality of sensors;

perform a second recognition process that is different from the first recognition process;

generate, based on reference information, first control information to control the second recognition process, wherein the reference information is based on the first recognition process;

output, as the reference information, specific information indicating an attention region in a dataset, wherein the attention region is extracted based on a feature amount from a specific layer associated with the first recognition process; and

generate second control information based on the specific information, to control the second.

11 . The information processing apparatus according to claim 10 , wherein a reading unit of the first sensor is one frame.

12 . The information processing apparatus according to claim 10 , wherein an output of the first recognition process includes the specific information indicating the attention region.

13 . An information processing method, comprising:

performing a first recognition process based on a first signal from a first sensor of a plurality of sensors;

performing a second recognition process based on a second signal from a second sensor of the plurality of sensors, wherein

a first reading unit of the first sensor is different from a second reading unit of the second sensor; and

generating control information for controlling the second recognition process, wherein

the control information is generated based on one of:

a dataset used for training in the first recognition process, or

output of the first recognition process.

14 . A non-transitory computer-readable medium having stored thereon computer-executable instructions that, when executed by a computer, cause the computer to execute operations, the operations comprising:

performing a first recognition process based on a first signal from a first sensor of a plurality of sensors;

performing a second recognition process based on a second signal from a second sensor of the plurality of sensors, wherein

a first reading unit of the first sensor is different from a second reading unit of the second sensor; and

generating control information for controlling the second recognition process, wherein

the control information is generated based on one of:

a dataset used for training in the first recognition process, or

output of the first recognition process.

15 . An information processing method, comprising:

performing a first recognition process based on a first signal from a first sensor of a plurality of sensors;

performing a second recognition process that is different from the first recognition process;

generating, based on reference information, first control information for controlling the second recognition process, wherein the reference information is based on the first recognition process;

outputting, as the reference information, specific information indicating an attention region in a dataset, wherein the attention region is extracted based on a feature amount from a specific layer associated with the first recognition process; and

generating second control information based on the specific information, for controlling the second recognition process.

16 . A non-transitory computer-readable medium having stored thereon computer-executable instructions that, when executed by a computer, cause the computer to execute operations, the operations comprising:

performing a first recognition process based on a first signal from a first sensor of a plurality of sensors;

performing a second recognition process that is different from the first recognition process;

generating, based on reference information, first control information for controlling the second recognition process, wherein the reference information is based on the first recognition process;

outputting, as the reference information, specific information indicating an attention region in a dataset, wherein the attention region is extracted based on a feature amount from a specific layer associated with the first recognition process; and

generating second control information based on the specific information, for controlling the second recognition process.