Search

EP-3499419-B1 - INFORMATION PROCESSING METHOD, INFORMATION PROCESSING APPARATUS, AND PROGRAM

EP3499419B1EP 3499419 B1EP3499419 B1EP 3499419B1EP-3499419-B1

Inventors

  • YAMAGUCHI, TAKUYA
  • ISHIKAWA, KAZUNOBU

Dates

Publication Date
20260506
Application Date
20181207

Claims (12)

  1. An information processing method, comprising the following executed using a computer: obtaining (S21) first sensor data (I1 to I4) obtained by an optical sensor (205) installed in a vehicle (200 200A), and at least one type of first traveling data (P1 to P7, D1 to D4) of the vehicle, the at least one type of first traveling data including any one of a location, time, weather, temperature, a traveling state, traveling speed, traveling control, and a driver attribute; associating (S22) the first sensor data and the at least one type of first traveling data with each other; characterized in that the method further comprises the step of: determining (S23) a degree of difference of the at least one type of first traveling data from the at least one type of a plurality of second traveling data each associated with a corresponding one of a plurality of second sensor data already selected as learning data for use in machine learning, the plurality of second sensor data each being associated with a corresponding one of the plurality of second traveling data in advance in storage included in the computer, the plurality of second traveling data being detected when the plurality of second sensor data are sensed; and, selecting (S24) the first sensor data as learning data according to the degree of difference, wherein in the selecting, the first sensor data associated with the at least one type of first traveling data is selected as the learning data, the at least one type of first traveling data having the degree of difference that is greater than or equal to a predetermined threshold value, and using the selected sensor data as the learning data for machine learning.
  2. The information processing method according to claim 1, further comprising: adding, to the plurality of second traveling data, the at least one type of first traveling data associated with the first sensor data selected as the learning data.
  3. The information processing method according to claim 1 or 2, wherein in the associating, the first sensor data and the at least one type of first traveling data obtained when the first sensor data is sensed are associated with each other.
  4. The information processing method according to any one of claims 1 to 3, wherein the at least one type of first traveling data comprises at least two types of first traveling data, and in the selecting, the first sensor data is selected as the learning data according to the degree of difference determined using the at least two types of first traveling data.
  5. The information processing method according to claim 4, wherein in the determining, the degree of difference is determined using a combination of the at least two types of first traveling data, and in the selecting, the first sensor data is selected as the learning data according to the degree of difference determined using the combination.
  6. The information processing method according to claim 4, wherein in the determining, the degree of difference is determined for each of the at least two types of first traveling data, using the at least two types of first traveling data, and in the selecting, the first sensor data is selected as the learning data according to the at least two degrees of difference determined respectively for the at least two types of first traveling data.
  7. The information processing method according to claim 6, wherein in the selecting, the at least two degrees of difference are integrated, and the first sensor data is selected as the learning data according to a result of the integration.
  8. The information processing method according to claim 6 or 7, wherein in the determining, the degree of difference is determined according to weight given to each of the at least two types of first traveling data, using the at least two types of first traveling data.
  9. The information processing method according to any one of claims 1 to 8, further comprising: normalizing the at least one type of first traveling data, wherein in the determining, the degree of difference is determined using the at least one type of first traveling data normalized.
  10. The information processing method according to any one of claims 1 to 9, further comprising: making a computation model for determining a degree of difference of the at least one type of first traveling data from the at least one type of one or more traveling data associated with one or more sensor data; and providing the computation model.
  11. An information processing apparatus comprising a processor (101); and a non-transitory recording medium storing (103) thereon a computer program, which when executed by the processor, causes the processor to perform the information processing method according to claim 1.
  12. A program for causing a computer to execute the information processing method according to claim 1.

Description

FIELD The present disclosure relates to information processing methods, information processing apparatuses, and programs for selecting sensor data as learning data. BACKGROUND Patent Literature (PTL) 1 discloses an error determination device that causes a selector or a learning agent to select information for learning from among image data each having an obtained detection value greater than or equal to a predetermined threshold value. CITATION LIST PATENT LITERATURE PTL 1: Japanese Unexamined Patent Application Publication No. 2016-173682PTL 2: European Patent Application No. EP 1 876 411 A1, disclosing an imaging position analysing method. SUMMARY TECHNICAL PROBLEM However, in PTL 1, since detection values of a specific error determination device are used, data effective for learning in the specific error determination device are selected. Accordingly, the technique of PTL 1 cannot always ensure a diversity of learning data for configurations or techniques of unspecified devices. In view of this, the present disclosure has an object to provide an information processing method, an information processing apparatus, and a program that can increase a diversity of learning data for configurations or techniques of unspecified devices. SOLUTIONS TO PROBLEM An information processing method according to the present invention is defined in the appended claims. Advantageous aspects are defined in the appended dependent claims. An information processing apparatus and program are also defined in the appended independent claims. It should be noted that these general or specific aspects may be implemented by a system, a device, an integrated circuit, a computer program, or a computer-readable recording medium such as a CD-ROM, and may be implemented by any combination of a system, a device, an integrated circuit, a computer program, and a recording medium. ADVANTAGEOUS EFFECT An information processing method, an information processing apparatus, and a program according to the present disclosure can increase a diversity of learning data for configurations or techniques of unspecified devices. BRIEF DESCRIPTION OF DRAWINGS FIG. 1 is a schematic view of an information processing system according to Embodiment 1.FIG. 2 is a block diagram illustrating an example of a hardware configuration of an information processing apparatus according to Embodiment 1.FIG. 3 is a block diagram illustrating an example of a hardware configuration of a vehicle according to Embodiment 1.FIG. 4 is a diagram illustrating an example of CAN (Controller Area Network) data.FIG. 5 is a block diagram illustrating an example of a functional configuration of the information processing system according to Embodiment 1.FIG. 6 is a diagram for illustrating an association between image data and traveling data by an association unit of the information processing apparatus.FIG. 7 is a table illustrating an example of combinations of associated image data and traveling data.FIG. 8 is a conceptual diagram unidimensionally representing a normal distribution of types of traveling data.FIG. 9 is a sequence diagram illustrating an example of operation in the information processing system.FIG. 10 is a block diagram illustrating an example of a functional configuration of a vehicle according to Embodiment 2.FIG. 11 is a flow chart illustrating an example of operation of the vehicle according to Embodiment 2. DESCRIPTION OF EMBODIMENTS (Underlying knowledge forming the basis of the present disclosure) In recent years, in the fields of automatic driving, security camera, robot, etc., objection detection using machine learning, such as deep learning for images captured by cameras, has been put into practical use. Such objection detection requires a large volume of teaching data for use in machine learning. For this purpose, myriad images captured by various cameras are collected, and teaching data are generated by humans giving correct interpretations to the collected images. However, since giving correct interpretations to images by humans is costly, it is undesirable to generate teaching data simply from all obtained myriad images. Besides, even if teaching data are generated by giving correct interpretations to all myriad images without considering costs, machine learning need be executed for the obtained large volume of teaching data. As a result, it takes more processing load and processing time to execute machine learning. Accordingly, for efficient execution of machine learning, it is necessary to select images effective for machine learning from among myriad images. Here, the myriad images used for machine learning need be made up of images captured in different situations, that is, diverse images. To put it differently, using images captured in different situations is more effective for achieving efficient machine learning than using images captured in similar situations. As described above, the error determination device narrows down obtained image data to ima