WO2022144992A1 - 情報処理装置、情報処理方法、及びコンピュータプログラム - Google Patents
情報処理装置、情報処理方法、及びコンピュータプログラム Download PDFInfo
- Publication number
- WO2022144992A1 WO2022144992A1 PCT/JP2020/049173 JP2020049173W WO2022144992A1 WO 2022144992 A1 WO2022144992 A1 WO 2022144992A1 JP 2020049173 W JP2020049173 W JP 2020049173W WO 2022144992 A1 WO2022144992 A1 WO 2022144992A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- likelihood ratio
- information processing
- series data
- processing apparatus
- Prior art date
Links
- 230000010365 information processing Effects 0.000 title claims abstract description 188
- 238000004590 computer program Methods 0.000 title claims description 18
- 238000003672 processing method Methods 0.000 title claims description 9
- 238000004364 calculation method Methods 0.000 claims abstract description 129
- 238000000034 method Methods 0.000 claims description 68
- 230000008569 process Effects 0.000 claims description 53
- 239000013598 vector Substances 0.000 description 183
- 230000008878 coupling Effects 0.000 description 38
- 238000010168 coupling process Methods 0.000 description 38
- 238000005859 coupling reaction Methods 0.000 description 38
- 238000010586 diagram Methods 0.000 description 27
- 230000000694 effects Effects 0.000 description 20
- 238000013500 data storage Methods 0.000 description 13
- 238000001514 detection method Methods 0.000 description 10
- 239000000284 extract Substances 0.000 description 10
- 230000006870 function Effects 0.000 description 10
- 238000013528 artificial neural network Methods 0.000 description 4
- 241001465754 Metazoa Species 0.000 description 3
- 230000001815 facial effect Effects 0.000 description 3
- 230000010354 integration Effects 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 241000270295 Serpentes Species 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000008685 targeting Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/90335—Query processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/26—Visual data mining; Browsing structured data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/532—Query formulation, e.g. graphical querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/63—Querying
- G06F16/632—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/68—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/683—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
Definitions
- This disclosure relates to, for example, the technical fields of information processing devices, information processing methods, and computer programs that process information related to classification.
- Patent Document 1 discloses that it is determined whether or not a person is a person by calculating a likelihood ratio based on the degree of similarity between the acquired biological information and a plurality of registered biological information.
- Patent Document 2 discloses that the log-likelihood ratio of the personal distribution and the distribution of others is obtained by using the collation score corresponding to the biological information.
- Patent Document 3 discloses that the score distribution is obtained by collating the input face image with the face images of all the registrants, and the registrant with the highest score is estimated to be the person himself / herself. ing.
- the integrated likelihood ratio is calculated from the first likelihood ratio and the second likelihood ratio, and the certainty of the facial posture candidates constituting the facial posture candidate group. It is disclosed to calculate the integrated likelihood representing.
- This disclosure is intended to improve the related techniques mentioned above.
- the series data is predetermined data based on an acquisition means for acquiring a plurality of elements included in the series data and at least two consecutive elements among the plurality of elements.
- a calculation means for calculating the likelihood ratio indicating the likelihood of being derived from the same target as the above, and a determination to determine whether or not the series data is derived from the same target as the predetermined data based on the likelihood ratio.
- the calculation means includes means, and the calculation means calculates the likelihood ratio by adding the degree of similarity or difference between the series data and the predetermined data.
- One aspect of the information processing method of the present disclosure is to acquire a plurality of elements included in the series data, and the series data is the same object as a predetermined data based on at least two consecutive elements among the plurality of elements.
- the likelihood ratio indicating the likelihood of origin is calculated, and based on the likelihood ratio, it is determined whether or not the series data is derived from the same object as the predetermined data, and the likelihood ratio is calculated. At the same time, the degree of similarity or difference between the series data and the predetermined data is added.
- One aspect of the computer program of this disclosure is to acquire a plurality of elements contained in the series data, and the series data is derived from the same object as a predetermined data based on at least two consecutive elements among the plurality of elements.
- the likelihood ratio indicating the likelihood of being At that time the computer is operated so as to take into account the degree of similarity or difference between the series data and the predetermined data.
- FIG. 1 is a block diagram showing a hardware configuration of the information processing apparatus according to the first embodiment.
- the information processing device 10 includes a processor 11, a RAM (Random Access Memory) 12, a ROM (Read Only Memory) 13, and a storage device 14.
- the information processing device 10 may further include an input device 15 and an output device 16.
- the processor 11, the RAM 12, the ROM 13, the storage device 14, the input device 15, and the output device 16 are connected via the data bus 17.
- Processor 11 reads a computer program.
- the processor 11 is configured to read a computer program stored in at least one of the RAM 12, the ROM 13, and the storage device 14.
- the processor 11 may read a computer program stored in a computer-readable recording medium using a recording medium reading device (not shown).
- the processor 11 may acquire (that is, read) a computer program from a device (not shown) arranged outside the information processing device 10 via a network interface.
- the processor 11 controls the RAM 12, the storage device 14, the input device 15, and the output device 16 by executing the read computer program.
- a functional block for executing the determination process using the likelihood ratio is realized in the processor 11.
- processor 11 a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an FPGA (field-programmable get array), a DSP (Demand-Side Platform), an ASIC, and an ASIC.
- processor 11 one of the above-mentioned examples may be used, or a plurality of processors 11 may be used in parallel.
- the RAM 12 temporarily stores the computer program executed by the processor 11.
- the RAM 12 temporarily stores data temporarily used by the processor 11 while the processor 11 is executing a computer program.
- the RAM 12 may be, for example, a D-RAM (Dynamic RAM).
- the ROM 13 stores a computer program executed by the processor 11.
- the ROM 13 may also store fixed data.
- the ROM 13 may be, for example, a P-ROM (Programmable ROM).
- the storage device 14 stores data stored in the information processing device 10 for a long period of time.
- the storage device 14 may operate as a temporary storage device of the processor 11.
- the storage device 14 may include, for example, at least one of a hard disk device, a magneto-optical disk device, an SSD (Solid State Drive), and a disk array device.
- the input device 15 is a device that receives an input instruction from the user of the information processing device 10.
- the input device 15 may include, for example, at least one of a keyboard, a mouse and a touch panel.
- the input device 15 may be a dedicated controller (operation terminal). Further, the input device 15 may include a terminal owned by the user (for example, a smartphone, a tablet terminal, or the like).
- the input device 15 may be a device capable of voice input including, for example, a microphone.
- the output device 16 is a device that outputs information about the information processing device 10 to the outside.
- the output device 16 may be a display device (for example, a display) capable of displaying information about the information processing device 10.
- the display device here may be a television monitor, a personal computer monitor, a smartphone monitor, a tablet terminal monitor, or another mobile terminal monitor.
- the display device may be a large monitor, a digital signage, or the like installed in various facilities such as a store.
- the output device 16 may be a device that outputs information in a format other than an image.
- the output device 16 may be a speaker that outputs information about the information processing device 10 by voice.
- FIG. 2 is a block diagram showing a functional configuration of the information processing apparatus according to the first embodiment.
- the information processing apparatus 10 includes a data acquisition unit 50, a likelihood ratio calculation unit 100, and a determination unit 200 as processing blocks for realizing the function. ing.
- Each of the data acquisition unit 50, the likelihood ratio calculation unit 100, and the determination unit 200 may be realized by, for example, the processor 11 (see FIG. 1) described above.
- the data acquisition unit 50 is configured to be able to acquire a plurality of elements included in the series data.
- the data acquisition unit 50 may acquire data directly from an arbitrary data acquisition device (for example, a camera, a microphone, etc.), or may read data previously acquired by the data acquisition device and stored in the storage or the like. It may be a thing.
- the data acquisition unit 50 may be configured to acquire data from each of the plurality of cameras.
- the elements of the series data acquired by the data acquisition unit 50 are configured to be output to the likelihood ratio calculation unit 100.
- the series data is data including a plurality of elements arranged in a predetermined order, and for example, time series data can be mentioned as an example. More specific examples of series data include, but are not limited to, video data and audio data.
- the likelihood ratio calculation unit 100 is configured to be able to calculate the likelihood ratio based on at least two consecutive elements among the plurality of elements acquired by the data acquisition unit 50.
- the "likelihood ratio" here is an index showing the likelihood that the series data is derived from the same target as the predetermined registered data.
- the likelihood ratio may be calculated, for example, as a log-likelihood cost (LLR: Log Likelihood Ratio). Further, the likelihood ratio may be calculated as an integrated likelihood ratio in which individual likelihood ratios calculated from two consecutive elements are integrated.
- the likelihood ratio calculation unit 100 can appropriately read information about the registered data from the registered data storage unit 300 that stores the registered data.
- the likelihood ratio calculation unit 100 according to the present embodiment is configured to be able to calculate the likelihood ratio, in particular, taking into account the degree of similarity or difference between the series data and the registered data. A specific method for calculating the likelihood ratio will be described in detail in other embodiments described later.
- the determination unit 200 determines whether or not the series data is derived from the same target as the registered data based on the likelihood ratio calculated by the likelihood ratio calculation unit 100. When the calculated likelihood ratio reaches a predetermined threshold value, the determination unit 200 may determine that the series data and the registered data are derived from the same target. Further, the determination unit 200 may determine that the series data is not derived from the same target as the registered data when the calculated likelihood ratio reaches another predetermined threshold value.
- FIG. 3 is a flowchart showing an operation flow of the information processing apparatus according to the first embodiment.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 calculates the likelihood ratio based on the acquired two or more elements. At this time, the likelihood ratio calculation unit 100 calculates the likelihood ratio by adding the degree of similarity or difference between the series data and the registered data (step S12). That is, the likelihood ratio is calculated as a value considering not only the acquired series data but also the registered data.
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated likelihood ratio (step S13).
- the determination unit 200 may output the determination result to a display or the like. Further, the determination unit 200 may output the determination result by voice via a speaker or the like.
- the determination unit 200 cannot determine whether the series data and the registered data are derived from the same object (for example, when the likelihood ratio does not reach the threshold value used for determination), the above-mentioned series of processes May be executed repeatedly. Specifically, the process of acquiring a new element from the series data and calculating the likelihood ratio in consideration of the newly acquired element may be repeated.
- FIG. 4 is a graph showing an example of the likelihood ratio calculated by the information processing apparatus according to the first embodiment.
- the likelihood ratio is calculated as a log-likelihood cost (LLR).
- This likelihood ratio gradually changes from the initial value by repeatedly executing the above-mentioned series of processes (see FIG. 3).
- the determination unit 200 determines using, for example, a first threshold value corresponding to a state in which the series data and the registered data are derived from the same object, and a second threshold value corresponding to a state in which the series data and the registered data are not derived from the same object. I do. Specifically, when the likelihood ratio reaches the first threshold value, the determination unit 200 determines that the series data and the registered data are derived from the same target. On the other hand, when the likelihood ratio reaches the second threshold value, the determination unit 200 determines that the series data and the registered data are not derived from the same target.
- the degree of similarity or the degree of difference between the series data and the registered data is taken into consideration when calculating the likelihood ratio. By doing so, it is possible to determine whether the series data and the registered data are derived from the same object by using the likelihood ratio calculated from the series data. More specifically, it is determined whether the input series data belongs to a class in which the series data and the registered data are derived from the same object, or a class in which the series data and the registered data are not derived from the same object. Can be done.
- the information processing apparatus 10 according to the second embodiment will be described with reference to FIGS. 5 to 7.
- the second embodiment explains specific examples of the series data and the registered data handled in the first embodiment described above.
- the apparatus configuration (see FIGS. 1 and 2) is described in the first embodiment. May be similar to. Therefore, in the following, the parts different from the first embodiment will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 5 is a diagram showing an example of image data handled by the information processing apparatus according to the second embodiment.
- FIG. 6 is a diagram showing an example of voice data handled by the information processing apparatus according to the second embodiment.
- the series data (hereinafter, may be appropriately referred to as “query”) and the registered data (hereinafter, appropriately referred to as “target”) handled by the information processing apparatus 10 according to the second embodiment may be referred to. ) May be image data, respectively.
- the series data and the registered data may be image data including a person (for example, face image data obtained by capturing a person's face).
- the series data and the registered data may be image data including an animal such as a snake or a dog, or an object such as a robot in addition to or instead of a person.
- the series data may be input as, for example, a plurality of time-series image data (that is, moving image data).
- the registered data may be registered as at least one image data. Further, the registered data may be moving image data or a 3D image. By using the image data including a person in this way, it is possible to determine whether or not the series data and the registered data are derived from the same person based on the calculated likelihood ratio.
- the series data and the registered data handled by the information processing apparatus 10 may be voice data, respectively. More specifically, the series data and the registered data may be data including voices emitted by a person. Alternatively, the series data and the registration data may be voice data emitted by an animal such as a snake or a dog, or an object such as a robot.
- the series data may be input as time-series voice data emitted by, for example, a person, an animal, or the like.
- the registered data may be registered as, for example, fixed-length voiceprint data.
- FIG. 7 is a flowchart showing an operation flow of the information processing apparatus according to the second embodiment.
- the data acquisition unit 50 first acquires the image data or the audio data included in the series data (step S21).
- the data acquisition unit 50 outputs the acquired image data or audio data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 extracts the feature amount from the acquired image data or audio data (step S22).
- the specific extraction method of the feature amount the existing technique can be appropriately adopted, and therefore detailed description thereof will be omitted here.
- the likelihood ratio calculation unit 100 calculates the likelihood ratio based on the extracted feature amount.
- the likelihood ratio calculation unit 100 calculates the likelihood ratio by adding the degree of similarity or difference between the series data and the registered data (step S23).
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same person based on the calculated likelihood ratio (step S24).
- the determination unit 200 may output the determination result to a display, a speaker, or the like. Further, the determination unit 200 may execute a predetermined process (for example, a process executed on condition of personal authentication) according to the determination result.
- the series data and the registered data include image data or audio data.
- the information processing device 10 according to the second embodiment can be applied to, for example, a device that performs face recognition.
- the voice data it can be determined whether or not the person who emitted the voice is the same person as the registered person. Therefore, the information processing device 10 according to the second embodiment can be applied to, for example, a device that performs voice authentication.
- the information processing apparatus 10 according to the third embodiment will be described with reference to FIGS. 8 to 10. It should be noted that the third embodiment may be the same as the first and second embodiments except for a part of the configuration and operation different from the above-mentioned first and second embodiments. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 8 is a block diagram showing a functional configuration of the information processing apparatus according to the third embodiment.
- the same components as those shown in FIG. 2 are designated by the same reference numerals.
- the information processing apparatus 10 includes a data acquisition unit 50, a likelihood ratio calculation unit 100, and a determination unit 200 as processing blocks for realizing the function. ing.
- the likelihood ratio calculation unit 100 according to the third embodiment is particularly configured to include a coupling unit 110.
- the coupling portion 110 may be realized by, for example, the processor 11 described above (see FIG. 1).
- the joining unit 110 can combine the feature vector extracted from the series data (hereinafter, appropriately referred to as “query vector”) and the feature vector extracted from the registered data (hereinafter, appropriately referred to as “target vector”). It is configured.
- the joining unit 110 joins the query vector and the target vector to generate a joining vector.
- the length of the join vector is the sum of the length of the query vector and the length of the target vector.
- the coupling vector generated by the coupling unit 110 is used to calculate the likelihood ratio.
- FIG. 9 is a flowchart showing an operation flow of the information processing apparatus according to the third embodiment.
- the same reference numerals are given to the same processes as those shown in FIG.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 extracts a feature vector (that is, a query vector) from the acquired elements of the series data (step S31).
- the joining unit 110 joins the query vector and the target vector to generate a joining vector (step S32).
- the target vector may be extracted in advance when registering the registration data or the like.
- the target vector may be stored in the registered data storage unit 300 together with the registered data, and the connecting unit 110 reads the target vector from the registered data storage unit 300 and combines it with the query vector.
- the target vector may be newly extracted from the registered data when the join vector is generated.
- the joining unit 110 reads the registered data from the registered data storage unit 300 and executes a process of extracting the target vector from the registered data. Then, the joining unit 110 joins the extracted target vector with the query vector.
- the likelihood ratio calculation unit 100 performs time series integration using the coupling vector and extracts the feature vector (step S33).
- the likelihood ratio calculation unit 100 inputs, for example, a coupling vector to an LSTM (Long Short Term Memory), and acquires a feature vector as its output.
- the LSTM here is an example, and the same processing may be executed using an arbitrary recursive neural network.
- a feature vector may be extracted using RNN (Recurrent Neural Network).
- the likelihood ratio calculation unit 100 converts the feature vector into a binary value by a transformation matrix (step S34). Specifically, the likelihood ratio calculation unit 100 converts the feature vector into a binary value indicating that the series data and the registered data are derived from the same object and that the series data and the registered data are not derived from the same target. do. Further, the likelihood ratio calculation unit 100 converts (that is, scales) the range of each of the two values into [0, 1].
- the likelihood ratio calculation unit 100 calculates the likelihood ratio from the converted value (step S35).
- the likelihood ratio calculated in this way is a value that takes into account the degree of similarity or difference between the series data and the registered data. Specifically, since the query vector extracted from the series data and the target vector extracted from the registered data are combined and the likelihood ratio is calculated based on the combined vector, the calculated likelihood ratio is the result. The degree of similarity or difference between the series data and the registered data is taken into consideration.
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated likelihood ratio (step S13).
- FIG. 10 is a conceptual diagram showing a specific operation example of the information processing apparatus according to the third embodiment.
- the operation example shown in FIG. 10 as described in the second embodiment (see FIGS. 5 to 7), it is determined whether or not the series data and the registered data are derived from the same person.
- the target vector is represented as t 1 .
- the joining unit 110 sequentially joins the query vector corresponding to each element and the target vector.
- the joining unit 110 first joins the query vector x 1 1 and the target vector t 1 to generate the joining vector t 1 x 1 1 . Then, the combined vector t 1 x 1 1 is input to the LSTM to extract the feature vector, and the likelihood ratio is calculated from the converted value of the feature vector. Subsequently, the coupling unit 110 combines the query vector x 1 2 and the target vector t 1 to generate the coupling vector t 1 x 1 2 . Then, the combined vector t 1 x 1 2 is input to the LSTM to extract the feature vector, and the likelihood ratio is calculated from the converted value of the feature vector.
- the coupling unit 110 combines the query vector x 1 M and the target vector t 1 to generate the coupling vector t 1 x 1 M. Then, the combined vector t 1 x 1 M is input to the LSTM to extract the feature vector, and the likelihood ratio is calculated from the converted value of the feature vector.
- the likelihood ratio calculated by the above processing, the personal threshold value (that is, the threshold value corresponding to the state in which the series data and the registered data are derived from the same person) and the other person's threshold value (that is, the series data and the registration) are shown.
- the determination is made by comparing the data with the threshold value corresponding to the state in which the data is not derived from the same person.
- the likelihood ratio gradually changes in the direction of the person's threshold value, and finally reaches the person's threshold value. Therefore, the determination unit 200 determines that the series data and the registered data are derived from the same person.
- the query vector and the target vector are combined to generate a coupling vector, and the likelihood ratio is calculated using the coupling vector. Processing is performed.
- the likelihood ratio calculated in this way is a value that takes into account the degree of similarity or difference between the series data and the registered data. Therefore, according to the information processing apparatus 10 according to the third embodiment, it is possible to determine whether the series data and the registered data are derived from the same object based on the calculated likelihood ratio.
- the information processing apparatus 10 according to the fourth embodiment will be described with reference to FIGS. 11 to 13. It should be noted that the fourth embodiment may be the same as the first to third embodiments except for a part of the configuration operation different from the above-mentioned first to third embodiments. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 11 is a block diagram showing a functional configuration of the information processing apparatus according to the fourth embodiment.
- the same components as those shown in FIG. 2 are designated by the same reference numerals.
- the information processing apparatus 10 includes a data acquisition unit 50, a likelihood ratio calculation unit 100, and a determination unit 200 as processing blocks for realizing the function. ing.
- the likelihood ratio calculation unit 100 according to the fourth embodiment is particularly configured to include a comparison unit 120.
- the comparison unit 120 may be realized by, for example, the processor 11 (see FIG. 1) described above.
- the comparison unit 120 is configured to be able to compare the feature vector generated by executing a predetermined process on the query vector extracted from the series data and the target vector. More specifically, the comparison unit 120 is configured to be able to calculate the degree of similarity between the feature amount vector generated from the query vector and the target vector. The comparison unit 120 may calculate the cosine similarity between the feature amount vector generated from the query vector and the target vector. However, the comparison unit 120 may calculate a similarity other than the cosine similarity. The similarity calculated by the comparison unit 120 is used to calculate the likelihood ratio.
- FIG. 12 is a flowchart showing a flow of operation of the information processing apparatus according to the fourth embodiment.
- the same reference numerals are given to the same processes as those shown in FIG.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 extracts a feature vector (that is, a query vector) from the acquired elements of the series data (step S41).
- a feature vector that is, a query vector
- the likelihood ratio calculation unit 100 performs time series integration using the query vector and extracts the feature vector (step S42).
- the likelihood ratio calculation unit 100 inputs, for example, a query vector to the LSTM, and acquires a feature vector as its output.
- the LSTM here is an example, and the same processing may be executed using an arbitrary recursive neural network.
- the feature vector may be extracted using RNN.
- the comparison unit 120 compares the feature amount vector extracted from the query vector with the target vector, and calculates the similarity of those vectors (step S43).
- the target vector may be extracted in advance when registering the registration data or the like.
- the target vector may be stored in the registered data storage unit 300 together with the registered data, and the comparison unit 120 reads the target vector from the registered data storage unit 300 and calculates the similarity.
- the target vector may be newly extracted from the registered data when calculating the similarity.
- the comparison unit 120 reads the registered data from the registered data storage unit 300 and executes a process of extracting the target vector from the registered data. Then, the comparison unit 120 compares the extracted target vector with the feature amount vector extracted from the query vector, and calculates the degree of similarity.
- the likelihood ratio calculation unit 100 converts the calculated range of similarity into the range of probability (step S44). For example, when the cosine similarity is calculated as the similarity, the likelihood ratio calculation unit 100 converts the cosine similarity range [-1,1] into the probability range [0,1].
- the likelihood ratio calculation unit 100 calculates the likelihood ratio from the converted value (step S45).
- the likelihood ratio calculated in this way is a value that takes into account the degree of similarity or difference between the series data and the registered data. Specifically, the similarity between the feature vector extracted from the query vector and the target vector is calculated, and the likelihood ratio is calculated based on the converted value of the similarity. As a result, the degree ratio takes into account the degree of similarity or difference between the series data and the registered data.
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated likelihood ratio (step S13).
- FIG. 13 is a conceptual diagram showing a specific operation example of the information processing apparatus according to the fourth embodiment.
- the operation example shown in FIG. 13 as described in the second embodiment (see FIGS. 5 to 7), it is determined whether or not the series data and the registered data are derived from the same person.
- the target vector is represented as t 1 .
- the comparison unit 120 sequentially calculates the similarity between the feature vector extracted from the query vector corresponding to each element and the target vector.
- the likelihood ratio calculation unit 100 first inputs the query vector x 11 into the LSTM and extracts the feature vector. Then, the comparison unit 120 calculates the degree of similarity between the feature vector extracted from the query vector x 11 and the target vector t 1 . After that, the likelihood ratio calculation unit 100 converts the calculated similarity to calculate the likelihood ratio. Subsequently, the likelihood ratio calculation unit 100 inputs the query vector x 12 into the LSTM and extracts the feature vector. Then, the comparison unit 120 calculates the degree of similarity between the feature vector extracted from the query vector x 12 and the target vector t 1 . After that, the likelihood ratio calculation unit 100 converts the calculated similarity to calculate the likelihood ratio.
- the likelihood ratio calculation unit 100 inputs the query vector x 1 M to the LSTM and extracts the feature vector. Then, the comparison unit 120 calculates the degree of similarity between the feature vector extracted from the query vector x 1 M and the target vector t 1 . After that, the likelihood ratio calculation unit 100 converts the calculated similarity to calculate the likelihood ratio.
- the determination is made by comparing the likelihood ratio calculated by the above-mentioned processing with the threshold value of the person and the threshold value of another person.
- the likelihood ratio gradually changes in the direction of the person's threshold value, and finally reaches the person's threshold value. Therefore, the determination unit 200 determines that the series data and the registered data are derived from the same person.
- a feature vector is extracted from the query vector, and the similarity is calculated by comparing the extracted feature vector with the target vector. .. Then, a process of calculating the likelihood ratio is performed based on the calculated similarity.
- the likelihood ratio calculated in this way is a value that takes into account the degree of similarity or difference between the series data and the registered data. Therefore, according to the information processing apparatus 10 according to the fourth embodiment, it is possible to determine whether the series data and the registered data are derived from the same object based on the calculated likelihood ratio.
- the information processing apparatus 10 according to the fifth embodiment will be described with reference to FIGS. 14 to 16. It should be noted that the fifth embodiment may be the same as the first to fourth embodiments except for a part of the configuration operation different from the above-mentioned first to fourth embodiments. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 14 is a block diagram showing a functional configuration of the information processing apparatus according to the fifth embodiment.
- the same components as those shown in FIG. 2 are designated by the same reference numerals.
- the information processing apparatus 10 includes a data acquisition unit 50, a likelihood ratio calculation unit 100, and a determination unit 200 as processing blocks for realizing the function. ing.
- the likelihood ratio calculation unit 100 according to the fifth embodiment is particularly configured to include a difference calculation unit 130.
- the difference calculation unit 130 may be realized by, for example, the processor 11 (see FIG. 1) described above.
- the difference calculation unit 130 is configured to be able to calculate the difference between the query vector extracted from the series data and the target vector extracted from the registered data.
- the difference calculation unit 130 calculates the difference vector as the difference between the query vector and the target vector.
- the difference vector calculated by the difference calculation unit 130 is used for calculating the likelihood ratio.
- FIG. 15 is a flowchart showing an operation flow of the information processing apparatus according to the fifth embodiment.
- the same reference numerals are given to the same processes as those shown in FIG.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 extracts a feature vector (that is, a query vector) from the acquired elements of the series data (step S51).
- the difference calculation unit 130 calculates the difference vector as the difference between the query vector and the target vector (step S52).
- the target vector may be extracted in advance when registering the registration data or the like.
- the target vector may be stored in the registered data storage unit 300 together with the registered data, and the difference calculation unit 130 reads the target vector from the registered data storage unit 300 and calculates the difference from the query vector.
- the target vector may be newly extracted from the registered data when calculating the difference vector.
- the difference calculation unit 130 reads the registered data from the registered data storage unit 300 and executes a process of extracting the target vector from the registered data. Then, the difference calculation unit 130 calculates the difference between the extracted target vector and the query vector.
- the likelihood ratio calculation unit 100 performs time series integration using the difference vector and extracts the feature vector (step S53).
- the likelihood ratio calculation unit 100 inputs, for example, a difference vector to the LSTM, and acquires a feature vector as its output.
- the LSTM here is an example, and the same processing may be executed using an arbitrary recursive neural network.
- the feature vector may be extracted using RNN.
- the likelihood ratio calculation unit 100 converts the feature vector into a binary value by a transformation matrix (step S54). Specifically, the likelihood ratio calculation unit 100 converts the feature vector into a binary value indicating that the series data and the registered data are derived from the same object and that the series data and the registered data are not derived from the same target. do. Further, the likelihood ratio calculation unit 100 converts (that is, scales) the range of each of the two values into [0, 1].
- the likelihood ratio calculation unit 100 calculates the likelihood ratio from the converted value (step S55).
- the likelihood ratio calculated in this way is a value that takes into account the degree of similarity or difference between the series data and the registered data. Specifically, since the difference between the query vector extracted from the series data and the target vector extracted from the registered data is calculated and the likelihood ratio is calculated based on the difference vector, the calculated likelihood ratio is As a result, the degree of similarity or difference between the series data and the registered data is added.
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated likelihood ratio (step S13).
- FIG. 16 is a conceptual diagram showing a specific operation example of the information processing apparatus according to the fifth embodiment.
- the operation example shown in FIG. 15 as described in the second embodiment (see FIGS. 5 to 7), it is determined whether or not the series data and the registered data are derived from the same person.
- the target vector is represented as t 1 .
- the difference calculation unit 130 sequentially calculates the difference between the query vector corresponding to each element and the target vector.
- the difference calculation unit 130 first calculates the difference between the query vector x 1 1 and the target vector t 1 to generate the difference vector t 1 ⁇ x 1 1 . Then, the difference vector t 1 ⁇ x 1 1 is input to the LSTM to extract the feature vector, and the likelihood ratio is calculated from the converted value of the feature vector. Subsequently, the difference calculation unit 130 calculates the difference between the query vector x 1 2 and the target vector t 1 to generate the difference vector t 1 ⁇ x 1 2 . Then, the difference vector t 1 ⁇ x 1 2 is input to the LSTM to extract the feature vector, and the likelihood ratio is calculated from the converted value of the feature vector.
- the difference calculation unit 130 calculates the difference between the query vector x 1 M and the target vector t 1 to generate the difference vector t 1 ⁇ x 1 M. .. Then, the difference vector t 1 ⁇ x 1 M is input to the LSTM to extract the feature vector, and the likelihood ratio is calculated from the converted value of the feature vector.
- the determination is made by comparing the likelihood ratio calculated by the above-mentioned processing with the threshold value of the person and the threshold value of another person.
- the likelihood ratio gradually changes in the direction of the person's threshold value, and finally reaches the person's threshold value. Therefore, the determination unit 200 determines that the series data and the registered data are derived from the same person.
- a difference vector is calculated as a difference between the query vector and the target vector, and the likelihood ratio is calculated using the difference vector. Processing is done.
- the likelihood ratio calculated in this way is a value that takes into account the degree of similarity or difference between the series data and the registered data. Therefore, according to the information processing apparatus 10 according to the fifth embodiment, it is possible to determine whether the series data and the registered data are derived from the same object based on the calculated likelihood ratio.
- the information processing apparatus 10 according to the sixth embodiment will be described with reference to FIGS. 17 and 18.
- the sixth embodiment selectively uses the above-mentioned third to fifth embodiments in combination, and the configuration and operation thereof may be substantially the same as those of the third to fifth embodiments. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 17 is a block diagram showing a functional configuration of the information processing apparatus according to the sixth embodiment.
- the same components as those shown in FIGS. 8, 11 and 14 are designated by the same reference numerals.
- the information processing apparatus 10 includes a data acquisition unit 50, a likelihood ratio calculation unit 100, and a determination unit 200 as processing blocks for realizing the function. ing.
- the likelihood ratio calculation unit 100 according to the sixth embodiment is particularly configured to include a coupling unit 110, a comparison unit 120, a difference calculation unit 130, and a selection unit 140. That is, the likelihood ratio calculation unit 100 according to the sixth embodiment is the coupling unit 110 described in the third embodiment, the comparison unit 120 described in the fourth embodiment, and the difference calculation unit 130 described in the fifth embodiment.
- a selection unit 140 is further provided.
- the selection unit 140 may be realized by, for example, the processor 11 (see FIG. 1) described above.
- the selection unit 140 is configured to be able to select which of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 is used when calculating the likelihood ratio. That is, the selection unit 140 uses the method of calculating the likelihood ratio using the coupling vector described in the third embodiment (see FIGS. 8 to 10) and the similarity ratio described in the fourth embodiment. (See FIGS. 11 to 13) and a method of calculating the likelihood ratio using the difference vector described in the fifth embodiment (see FIGS. 14 to 16). It is possible to select whether to calculate the likelihood ratio.
- the selection unit 140 selects which of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 is to be used based on the condition information acquired in advance.
- the condition information is information for determining which of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 is optimal for calculating the likelihood ratio.
- the condition information may be, for example, information regarding the type of series data or registered data. For example, information indicating that the series data and the registered data are image data, and information indicating that the series data and the registered data are audio data may be acquired as conditional information. Further, the condition information may be, for example, information regarding the quality of series data or registered data. For example, when the series data and the registered data are image data or audio data, information indicating the sharpness may be acquired as condition information.
- condition information may be information about the environment in which the series data or the registration data is acquired.
- the series data and registered data are image data or audio data
- information about the place where the data was acquired and the surrounding environment information about the camera used to acquire the image data
- audio data information about the microphone used to acquire the data.
- Information about the microphone used to acquire the data may be acquired as conditional information.
- FIG. 18 is a flowchart showing a flow of operation of the information processing apparatus according to the sixth embodiment.
- the selection unit 140 first acquires the condition information (step S61). Then, the selection unit 140 selects which of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 is used when calculating the likelihood ratio based on the condition information (step S62).
- the information processing apparatus 10 uses a method selected by the selection unit 140 to perform a determination process (that is, a process of calculating a likelihood ratio to determine whether the series data and the registered data are derived from the same object). Is executed (step S63). Specifically, when the selection unit 140 selects the coupling unit 110, the determination process (see FIG. 9) described in the third embodiment is executed. When the selection unit 140 selects the comparison unit 120, the determination process (see FIG. 12) described in the fourth embodiment is executed. When the selection unit 140 selects the difference calculation unit 130, the determination process (see FIG. 15) described in the fifth embodiment is executed.
- a determination process that is, a process of calculating a likelihood ratio to determine whether the series data and the registered data are derived from the same object.
- the condition information is acquired and the likelihood ratio calculation method is selected before the determination process is executed.
- the condition information is acquired during the determination process.
- the method of calculating the likelihood ratio may be selected. For example, after the data acquisition unit acquires an element from the series data (that is, after step S11), the condition information may be acquired and the likelihood ratio calculation method may be selected. Further, the condition information may be acquired immediately before the coupling vector is generated by the coupling unit 110 (that is, immediately before step S32 in FIG. 9), and the likelihood ratio calculation method may be selected. Conditional information may be acquired and the likelihood ratio calculation method may be selected immediately before the similarity is calculated by the comparison unit 120 (that is, immediately before step S43 in FIG. 12). Conditional information may be acquired immediately before the difference vector is calculated by the difference calculation unit 130 (that is, immediately before step S52 in FIG. 15), and the likelihood ratio calculation method may be selected.
- the likelihood ratio is calculated by the selection unit 140, the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 are used. Which one to use is selected. By doing so, the likelihood ratio can be calculated by selecting the optimum method from the three units of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130. Therefore, it is possible to accurately determine whether the series data and the registered data are derived from the same object.
- FIG. 19 is a block diagram showing a functional configuration of the information processing apparatus according to the modified example of the sixth embodiment.
- the same components as those shown in FIGS. 8, 11 and 14 are designated by the same reference numerals.
- the information processing apparatus 10 has a data acquisition unit 50, a likelihood ratio calculation unit 100, and a determination unit 200 as processing blocks for realizing the function. And have.
- the likelihood ratio calculation unit 100 according to the modified example of the sixth embodiment is particularly configured to include a coupling unit 110, a comparison unit 120, a difference calculation unit 130, and an operation detection unit 150. That is, the likelihood ratio calculation unit 100 according to the modified example of the sixth embodiment is the coupling unit 110 described in the third embodiment, the comparison unit 120 described in the fourth embodiment, and the difference described in the fifth embodiment.
- a calculation unit 130 is provided, and an operation detection unit 150 is further provided in addition to the calculation unit 130.
- the operation detection unit 150 may be realized by, for example, the processor 11 (see FIG. 1) described above.
- the operation detection unit 150 is configured to be able to detect an operation by the user. Specifically, the operation detection unit 150 is an operation of selecting which of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 is used when calculating the likelihood ratio (hereinafter, “selection operation” as appropriate). Is configured to be detectable. In other words, the operation detection unit 150 uses the method of calculating the likelihood ratio using the coupling vector described in the third embodiment (see FIGS. 8 to 10) and the similarity described in the fourth embodiment. Which of the methods for calculating the likelihood ratio (see FIGS. 11 to 13) and the method for calculating the likelihood ratio using the difference vector described in the fifth embodiment (see FIGS. 14 to 16) can be used. It is possible to detect an operation of selecting whether to calculate the likelihood ratio using the product. The operation detection unit 150 may detect, for example, a user's selection operation by the input device 15 (see FIG. 1).
- FIG. 20 is a flowchart showing the operation flow of the information processing apparatus according to the modified example of the sixth embodiment.
- the operation detection unit 150 first detects the selection operation by the user (step S65). Then, the operation detection unit 150 selects which of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 is used when calculating the likelihood ratio based on the detected selection operation (step S66). ..
- a notification or the like urging the user to perform the selection operation may be given.
- a speaker or the like may be used to notify the user to perform the selection operation by voice.
- a display or the like may be used to notify the user to perform the selection operation on the screen display.
- the user may touch the screen to perform a selection operation. For example, even if three areas corresponding to each of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 are displayed on the screen, and an operation in which the user touches any of the areas is detected as a selection operation. good.
- the information processing apparatus 10 uses a method corresponding to the user's selection operation to perform a determination process (that is, a process of calculating a likelihood ratio to determine whether the series data and the registered data are derived from the same object. ) Is executed (step S67). Specifically, when the user performs an operation of selecting the coupling portion 110, the determination process (see FIG. 9) described in the third embodiment is executed. When the user performs an operation of selecting the comparison unit 120, the determination process (see FIG. 12) described in the fourth embodiment is executed. When the user performs an operation of selecting the difference calculation unit 130, the determination process (see FIG. 15) described in the fifth embodiment is executed.
- a determination process that is, a process of calculating a likelihood ratio to determine whether the series data and the registered data are derived from the same object.
- the selection operation may be detected in the middle of the determination process.
- the selection operation may be detected after the data acquisition unit acquires an element from the series data (that is, after step S11).
- the selection operation may be detected immediately before the connection vector is generated by the connection unit 110 (that is, immediately before step S32 in FIG. 9).
- the selection operation may be detected immediately before the similarity is calculated by the comparison unit 120 (that is, immediately before step S43 in FIG. 12).
- the selection operation may be detected immediately before the difference vector is calculated by the difference calculation unit 130 (that is, immediately before step S52 in FIG. 15).
- the likelihood ratio is calculated according to the operation of the user, the coupling unit 110, the comparison unit 120, and the like. And which of the difference calculation unit 130 is used is selected. By doing so, the likelihood ratio can be calculated by selecting the optimum method from the three units of the coupling unit 110, the comparison unit 120, and the difference calculation unit 130. Therefore, it is possible to accurately determine whether the series data and the registered data are derived from the same object.
- the coupling unit 110, the comparison unit 120, and the difference calculation unit 130 are selectively used, but the coupling unit 110, Two of the comparison unit 120 and the difference calculation unit 130 may be selectively used.
- the coupling unit 110 and the comparison unit 120 may be selectively used.
- the coupling unit 110 and the difference calculation unit 130 may be selectively used.
- the comparison unit 120 and the difference calculation unit 130 may be selectively used.
- the information processing apparatus 10 according to the seventh embodiment will be described with reference to FIGS. 21 and 22. It should be noted that the seventh embodiment differs from the above-mentioned first to sixth embodiments only in a part of the configuration and operation, and other parts may be the same as those of the first to sixth embodiments. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 21 is a diagram conceptually showing a determination targeting a plurality of registered data in the information processing apparatus according to the seventh embodiment.
- a plurality of registered data are subject to determination with respect to the input series data. That is, it is configured to determine whether or not one query is derived from the same target as a plurality of targets.
- a plurality of registered data it is assumed that facial images of a plurality of persons are registered as a plurality of registered data. In this case, it is determined which of the person in the image input as the series data and the plurality of registered people is the same person.
- the likelihood ratio considering the series data and each of the plurality of registered data may be calculated.
- a plurality of likelihood ratios may be calculated in consideration of the degree of similarity or difference between the series data and each of the plurality of registered data.
- FIG. 22 is a flowchart showing a flow of operation of the information processing apparatus according to the seventh embodiment.
- the same reference numerals are given to the same processes as those shown in FIG.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 calculates the likelihood ratio based on the acquired two or more elements. At this time, the likelihood ratio calculation unit 100 calculates a plurality of likelihood ratios by adding the degree of similarity or difference between the series data and the plurality of registered data (step S71). That is, the likelihood ratio calculates a plurality of likelihood ratios according to the number of determination targets (that is, the number of registered data).
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated plurality of likelihood ratios (step S72). For example, when the first likelihood ratio reaches the person threshold value, the determination unit 200 determines that the series data and the registered data corresponding to the first likelihood ratio are derived from the same person. Similarly, when the second likelihood ratio reaches the person threshold value, the determination unit 200 determines that the series data and the registered data corresponding to the second likelihood ratio are derived from the same person.
- a plurality of likelihood ratios are taken into consideration in consideration of the degree of similarity or difference between the series data and each of the plurality of registered data. Is calculated. In this way, even when there are a plurality of registered data (that is, when there are a plurality of determination targets), it is possible to determine which of the series data and which registered data are derived from the same target. can.
- the information processing apparatus 10 according to the eighth embodiment will be described with reference to FIGS. 23 and 24. It should be noted that the eighth embodiment is different from the above-mentioned seventh embodiment only in a part of the configuration and operation, and other parts may be the same as the seventh embodiment. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 23 is a block diagram showing a functional configuration of the information processing apparatus according to the eighth embodiment.
- the same components as those shown in FIG. 2 are designated by the same reference numerals.
- the information processing apparatus 10 has a data acquisition unit 50, a likelihood ratio calculation unit 100, a determination unit 200, and an output as a processing block for realizing the function. It is provided with a unit 400. That is, the information processing apparatus 10 according to the eighth embodiment is configured to further include an output unit 400 in addition to the configuration of the first embodiment (see FIG. 2).
- the output unit 400 may be realized by, for example, the processor 11 (see FIG. 1) described above. Further, the output unit 400 may be configured to include the above-mentioned output device (see FIG. 1).
- the output unit 400 is configured to be able to output the determination result by the determination unit 200. That is, the output unit 400 is configured to be able to output a determination result as to whether or not the series data and the registered data are derived from the same target. Further, the output unit 400 is configured to be capable of outputting not only one registered data derived from the same target as the series data but also a plurality of registered data likely to be derived from the same target as the series data as a determination result. ing.
- the output unit 400 may output the determination result to a display or the like. Further, the output unit 400 may output the determination result by voice via a speaker or the like.
- FIG. 24 is a flowchart showing the operation flow of the information processing apparatus according to the eighth embodiment.
- the same reference numerals are given to the same processes as those shown in FIG. 22.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the likelihood ratio calculation unit 100 calculates the likelihood ratio based on the acquired two or more elements. At this time, the likelihood ratio calculation unit 100 calculates a plurality of likelihood ratios by adding the degree of similarity or difference between the series data and the plurality of registered data (step S71). That is, the likelihood ratio calculates a plurality of likelihood ratios according to the number of determination targets (that is, the number of registered data).
- the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated plurality of likelihood ratios (step S72). Then, the output unit 400 outputs information regarding one or a plurality of registered data as the determination result by the determination unit 200 (step S81).
- the output unit 400 may output only the information regarding one registered data corresponding to the likelihood ratio as the determination result.
- the output unit 400 has a plurality of likelihood ratios corresponding to the plurality of likelihood ratios. Information on the registered data may be output as a determination result.
- the output unit 400 when there is no likelihood ratio that has reached the person's threshold value among the plurality of likelihood ratios corresponding to each of the plurality of registered data, the output unit 400 has a predetermined number in the order in which the final value is closer to the person's threshold value.
- the likelihood ratio may be selected, and information about a plurality of registered data corresponding to the selected likelihood ratio may be output as a determination result.
- the output unit 400 selects a plurality of likelihood ratios whose final value exceeds the selection threshold lower than the principal threshold, and selects the likelihood ratio. Information about a plurality of registered data corresponding to the above may be output as a determination result.
- one registered data derived from the same target as the series data, or a plurality of data having a high possibility of being derived from the same target as the series data is configured to be able to output the registration data of. By doing so, even in the case where the registered data derived from the same target as the series data cannot be accurately narrowed down to one, a plurality of candidates can be output as the determination result.
- ⁇ 9th embodiment> The information processing apparatus 10 according to the ninth embodiment will be described with reference to FIGS. 25 and 26. It should be noted that the ninth embodiment is different from the above-mentioned seventh embodiment only in a part of the configuration and operation, and other parts may be the same as the seventh embodiment. Therefore, in the following, the parts different from each of the above-described embodiments will be described in detail, and the description of other overlapping parts will be omitted as appropriate.
- FIG. 25 is a block diagram showing a functional configuration of the information processing apparatus according to the ninth embodiment.
- the same components as those shown in FIG. 2 are designated by the same reference numerals.
- the information processing apparatus 10 determines the data acquisition unit 50, the likelihood ratio calculation unit 100, and the determination unit 200 as processing blocks for realizing the function. It is equipped with a target limited unit 500. That is, the information processing apparatus 10 according to the ninth embodiment is configured to further include a determination target limiting unit 500 in addition to the configuration of the first embodiment (see FIG. 2).
- the determination target limiting unit 500 may be realized by, for example, the processor 11 (see FIG. 1) described above.
- the determination target limiting unit 500 is configured to be able to limit the determination target by executing a narrowing process on a plurality of registered data stored in the registered data storage unit 300. That is, the determination target limiting unit 500 is configured to be able to execute a process for reducing the number of determination targets.
- the narrowing down process executed by the determination target limiting unit 500 may be a face authentication process having a low processing load. The face recognition process in this case does not have to be so accurate (for example, it may have a slightly higher tolerance for others).
- the determination target is only the registration data that may be derived from the same person as the series data. It can be narrowed down to (that is, a smaller number of registered data than all stored registered data).
- FIG. 26 is a flowchart showing the operation flow of the information processing apparatus according to the ninth embodiment.
- the same reference numerals are given to the same processes as those shown in FIG. 22.
- the data acquisition unit 50 first acquires the elements included in the series data (step S11).
- the data acquisition unit 50 outputs the elements of the acquired series data to the likelihood ratio calculation unit 100.
- the determination target limiting unit 500 executes the narrowing process using the acquired elements of the series data to limit the number of determination targets (step S91).
- the subsequent processing will be executed only for the limited determination target. That is, the execution is performed not for all the registered data stored in the registered data storage unit 300, but for the registered data whose number has been reduced by the narrowing process.
- the likelihood ratio calculation unit 100 calculates the likelihood ratio based on the acquired two or more elements. At this time, the likelihood ratio calculation unit 100 calculates a plurality of likelihood ratios by adding the degree of similarity or difference between the series data and the limited registered data (step S92). Then, the determination unit 200 determines whether or not the series data and the registered data are derived from the same target based on the calculated plurality of likelihood ratios (step S93).
- the registration data to be determined is limited by the narrowing process (that is, the number of determination targets is reduced). By doing so, even if the original number of determination targets is enormous, it is possible to reduce the number of determination objects for which the likelihood ratio is actually calculated. Therefore, the processing load of the information processing apparatus 10 and the time required for processing can be effectively suppressed.
- Each embodiment also implements a processing method in which a program for operating the configuration of the embodiment is recorded on a recording medium so as to realize the functions of the above-described embodiments, the program recorded on the recording medium is read out as a code, and the program is executed by a computer. Included in the category of morphology. That is, a computer-readable recording medium is also included in the scope of each embodiment. Further, not only the recording medium on which the above-mentioned program is recorded but also the program itself is included in each embodiment.
- the recording medium for example, a floppy (registered trademark) disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a magnetic tape, a non-volatile memory card, or a ROM can be used.
- a floppy (registered trademark) disk for example, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a magnetic tape, a non-volatile memory card, or a ROM
- the program recorded on the recording medium is executed by itself, but also the program that operates on the OS and executes the process in cooperation with other software and the function of the expansion board is also an embodiment. Is included in the category of.
- the series data is the same as the predetermined data based on the acquisition means for acquiring the plurality of elements included in the series data and at least two consecutive elements among the plurality of elements.
- the calculation means is an information processing apparatus characterized in that the likelihood ratio is calculated by adding the degree of similarity or difference between the series data and the predetermined data.
- the calculation means is the first with respect to the combined feature amount obtained by combining the first feature amount extracted from the series data and the second feature amount extracted from the predetermined data.
- Addendum 1 characterized in that the degree of similarity or difference between the series data and the predetermined data is added by performing the processing and calculating the likelihood ratio based on the processing result of the first processing.
- Appendix 3 In the information processing apparatus according to Appendix 3, the calculation means performs a second process on the first feature amount extracted from the series data, and the first feature amount after the second process and the predetermined data.
- Addendum 1 characterized in that the degree of similarity or difference between the series data and the predetermined data is calculated by calculating the likelihood ratio based on the comparison result with the second feature amount extracted from.
- the calculation means is the first with respect to the difference feature amount which is the difference between the first feature amount extracted from the series data and the second feature amount extracted from the predetermined data.
- Addendum 1 characterized in that the degree of similarity or difference between the series data and the predetermined data is calculated by performing the three processes and calculating the likelihood ratio based on the processing result of the third process. It is an information processing apparatus according to.
- the calculation means calculates a plurality of the likelihood ratios corresponding to each of the plurality of predetermined data, and the determination means is based on the plurality of the likelihood ratios.
- the information processing apparatus according to any one of Supplementary note 1 to 4, wherein it is determined whether or not the series data is derived from the same target as any of the plurality of predetermined data.
- Appendix 6 determines that the determination means is likely to be one predetermined data determined to be derived from the same target as the series data, or to be derived from the same target as the series data.
- Appendix 7 The information processing apparatus according to Appendix 7 is further provided with limiting means for narrowing down a plurality of the predetermined data and limiting the number of the predetermined data to be determined by the determination means to be reduced.
- the series data includes at least one of the target image data and the voice data, and the calculation means is likely that the series data is derived from the same target as the predetermined data.
- Appendix 9 In the information processing method described in Appendix 9, a plurality of elements included in the series data are acquired, and the series data is derived from the same object as the predetermined data based on at least two consecutive elements among the plurality of elements.
- the likelihood ratio indicating the plausibility of a certain thing
- it is an information processing method characterized by adding the degree of similarity or the degree of difference between the series data and the predetermined data.
- Appendix 10 The computer program according to Appendix 10 acquires a plurality of elements included in the series data, and the series data is derived from the same object as the predetermined data based on at least two consecutive elements among the plurality of elements.
- the likelihood ratio indicating the likelihood of the fact
- a computer program characterized in that a computer is operated so as to take into account the degree of similarity or difference between the series data and the predetermined data.
- Appendix 11 The recording medium described in Appendix 11 is a recording medium characterized in that the computer program described in Appendix 10 is recorded.
- Information processing device 11 10 Information processing device 11
- Processor 14 Storage device 50
- Data acquisition unit 100
- Probability ratio calculation unit 110 Coupling unit 120 Comparison unit 130 Difference calculation unit 140
- Selection unit 150
- Operation detection unit 200
- Judgment unit 300
- Registered data storage unit 400
- Output unit 500 Judgment target Limited section
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Library & Information Science (AREA)
- Multimedia (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
Description
第1実施形態に係る情報処理装置について、図1から図4を参照して説明する。
まず、図1を参照しながら、第1実施形態に係る情報処理装置のハードウェア構成について説明する。図1は、第1実施形態に係る情報処理装置のハードウェア構成を示すブロック図である。
次に、図2を参照しながら、第1実施形態に係る情報処理装置10の機能的構成について説明する。図2は、第1実施形態に係る情報処理装置の機能的構成を示すブロック図である。
次に、図3を参照しながら、第1実施形態に係る情報処理装置10の動作の流れについて説明する。図3は、第1実施形態に係る情報処理装置の動作の流れを示すフローチャートである。
次に、図4を参照しながら、第1実施形態に係る情報処理装置10による具体的な判定例について説明する。図4は、第1実施形態に係る情報処理装置で算出される尤度比の一例を示すグラフである。なお、図4の例では、尤度比が対数尤度費(LLR)として算出されているものとする。
次に、第1実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第2実施形態に係る情報処理装置10について、図5から図7を参照して説明する。なお、第2実施形態は、上述した第1実施形態で扱われる系列データ及び登録データの具体例を説明するものであり、例えば装置構成(図1及び図2参照)については、第1実施形態と同様であってよい。このため、以下では、第1実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図5及び図6を参照しながら、第2実施形態に係る情報処理装置10で扱われる系列データ及び登録データについて説明する。図5は、第2実施形態に係る情報処理装置で扱われる画像データの一例を示す図である。図6は、第2実施形態に係る情報処理装置で扱われる音声データの一例を示す図である。
次に、図7を参照しながら、第2実施形態に係る情報処理装置10の動作の流れについて説明する。図7は、第2実施形態に係る情報処理装置の動作の流れを示すフローチャートである。
次に、第2実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第3実施形態に係る情報処理装置10について、図8から図10を参照して説明する。なお、第3実施形態は、上述した第1及び第2実施形態と一部の構成及び動作が異なるのみで、その他の部分については第1及び第2実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図8を参照しながら、第3実施形態に係る情報処理装置10の機能的構成について説明する。図8は、第3実施形態に係る情報処理装置の機能的構成を示すブロック図である。なお、図8では、図2に示した構成要素と同様のものに同一の符号を付している。
次に、図9を参照しながら、第3実施形態に係る情報処理装置10の動作の流れについて説明する。図9は、第3実施形態に係る情報処理装置の動作の流れを示すフローチャートである。なお、図9では、図3で示した処理と同様の処理に同一の符号を付している。
次に、図10を参照しながら、第3実施形態に係る情報処理装置10の具体的な動作例を説明する。図10は、第3実施形態に係る情報処理装置の具体的な動作例を示す概念図である。なお、図10に示す動作例では、第2実施形態で説明したように(図5から図7を参照)、系列データと登録データとが同一人物由来であるか否かを判定している。
次に、第3実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第4実施形態に係る情報処理装置10について、図11から図13を参照して説明する。なお、第4実施形態は、上述した第1から第3実施形態と一部の構成動作が異なるのみで、その他の部分については第1から第3実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図11を参照しながら、第4実施形態に係る情報処理装置10の機能的構成について説明する。図11は、第4実施形態に係る情報処理装置の機能的構成を示すブロック図である。なお、図11では、図2に示した構成要素と同様のものに同一の符号を付している。
次に、図12を参照しながら、第4実施形態に係る情報処理装置10の動作の流れについて説明する。図12は、第4実施形態に係る情報処理装置の動作の流れを示すフローチャートである。なお、図12では、図3で示した処理と同様の処理に同一の符号を付している。
次に、図13を参照しながら、第4実施形態に係る情報処理装置10の具体的な動作例を説明する。図13は、第4実施形態に係る情報処理装置の具体的な動作例を示す概念図である。なお、図13に示す動作例では、第2実施形態で説明したように(図5から図7を参照)、系列データと登録データとが同一人物由来であるか否かを判定している。
次に、第4実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第5実施形態に係る情報処理装置10について、図14から図16を参照して説明する。なお、第5実施形態は、上述した第1から第4実施形態と一部の構成動作が異なるのみで、その他の部分については第1から第4実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図14を参照しながら、第5実施形態に係る情報処理装置10の機能的構成について説明する。図14は、第5実施形態に係る情報処理装置の機能的構成を示すブロック図である。なお、図14では、図2に示した構成要素と同様のものに同一の符号を付している。
次に、図15を参照しながら、第5実施形態に係る情報処理装置10の動作の流れについて説明する。図15は、第5実施形態に係る情報処理装置の動作の流れを示すフローチャートである。なお、図15では、図3で示した処理と同様の処理に同一の符号を付している。
次に、図16を参照しながら、第5実施形態に係る情報処理装置10の具体的な動作例を説明する。図16は、第5実施形態に係る情報処理装置の具体的な動作例を示す概念図である。なお、図15に示す動作例では、第2実施形態で説明したように(図5から図7を参照)、系列データと登録データとが同一人物由来であるか否かを判定している。
次に、第5実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第6実施形態に係る情報処理装置10について、図17及び図18を参照して説明する。なお、第6実施形態は、上述した第3から第5実施形態を組み合わせて選択的に利用するものであり、その構成や動作は概ね第3から第5実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図17を参照しながら、第6実施形態に係る情報処理装置10の機能的構成について説明する。図17は、第6実施形態に係る情報処理装置の機能的構成を示すブロック図である。なお、図17では、図8、図11及び図14に示した構成要素と同様のものに同一の符号を付している。
次に、図18を参照しながら、第6実施形態に係る情報処理装置10の動作の流れについて説明する。図18は、第6実施形態に係る情報処理装置の動作の流れを示すフローチャートである。
次に、第6実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第6実施形態の変形例に係る情報処理装置10について、図19及び図20を参照して説明する。なお、第6実施形態の変形例は、上述した第6実施形態と一部の構成及び動作が異なるのみであり、その他の部分については第6実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図19を参照しながら、第6実施形態の変形例に係る情報処理装置10の機能的構成について説明する。図19は、第6実施形態の変形例に係る情報処理装置の機能的構成を示すブロック図である。なお、図19では、図8、図11及び図14に示した構成要素と同様のものに同一の符号を付している。
次に、図20を参照しながら、第6実施形態の変形例に係る情報処理装置10の動作の流れについて説明する。図20は、第6実施形態の変形例に係る情報処理装置の動作の流れを示すフローチャートである。
次に、第6実施形態の変形例に係る情報処理装置10によって得られる技術的効果について説明する。
第7実施形態に係る情報処理装置10について、図21及び図22を参照して説明する。なお、第7実施形態は、上述した第1から第6実施形態と一部の構成及び動作が異なるのみであり、その他の部分については第1から第6実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図21を参照しながら、第7実施形態に係る情報処理装置10が実行する、複数の登録データを対象とする判定について説明する。図21は、第7実施形態に係る情報処理装置における複数の登録データを対象とする判定を概念的に示す図である。
次に、図22を参照しながら、第7実施形態に係る情報処理装置10の動作の流れについて説明する。図22は、第7実施形態に係る情報処理装置の動作の流れを示すフローチャートである。なお、図22では、図3で示した処理と同様の処理に同一の符号を付している。
次に、第7実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第8実施形態に係る情報処理装置10について、図23及び図24を参照して説明する。なお、第8実施形態は、上述した第7実施形態と一部の構成及び動作が異なるのみであり、その他の部分については第7実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図23を参照しながら、第8実施形態に係る情報処理装置10の機能的構成について説明する。図23は、第8実施形態に係る情報処理装置の機能的構成を示すブロック図である。なお、図23では、図2に示した構成要素と同様のものに同一の符号を付している。
次に、図24を参照しながら、第8実施形態に係る情報処理装置10の動作の流れについて説明する。図24は、第8実施形態に係る情報処理装置の動作の流れを示すフローチャートである。なお、図24では、図22で示した処理と同様の処理に同一の符号を付している。
次に、第8実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
第9実施形態に係る情報処理装置10について、図25及び図26を参照して説明する。なお、第9実施形態は、上述した第7実施形態と一部の構成及び動作が異なるのみであり、その他の部分については第7実施形態と同様であってよい。このため、以下では、すでに説明した各実施形態と異なる部分について詳しく説明し、他の重複する部分については適宜説明を省略するものとする。
まず、図25を参照しながら、第9実施形態に係る情報処理装置10の機能的構成について説明する。図25は、第9実施形態に係る情報処理装置の機能的構成を示すブロック図である。なお、図25では、図2に示した構成要素と同様のものに同一の符号を付している。
次に、図26を参照しながら、第9実施形態に係る情報処理装置10の動作の流れについて説明する。図26は、第9実施形態に係る情報処理装置の動作の流れを示すフローチャートである。なお、図26では、図22で示した処理と同様の処理に同一の符号を付している。
次に、第9実施形態に係る情報処理装置10によって得られる技術的効果について説明する。
以上説明した実施形態に関して、更に以下の付記のようにも記載されうるが、以下には限られない。
付記1に記載の情報処理装置は、系列データに含まれる複数の要素を取得する取得手段と、前記複数の要素のうち少なくとも2つの連続する要素に基づいて、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出する算出手段と、前記尤度比に基づいて、前記系列データが所定のデータと同一対象由来であるか否かを判定する判定手段とを備え、前記算出手段は、前記系列データと前記所定のデータとの類似度又は相違度を加味して、前記尤度比を算出することを特徴とする情報処理装置である。
付記2に記載の情報処理装置は、前記算出手段は、前記系列データから抽出した第1特徴量と、前記所定のデータから抽出した第2特徴量とを結合した結合特徴量に対して第1処理を行い、前記第1処理の処理結果に基づいて前記尤度比を算出することで、前記系列データと前記所定のデータとの類似度又は相違度を加味することを特徴とする付記1に記載の情報処理装置である。
付記3に記載の情報処理装置は、前記算出手段は、前記系列データから抽出した第1特徴量に対して第2処理を行い、前記第2処理後の前記第1特徴量と前記所定のデータから抽出した第2特徴量との比較結果に基づいて前記尤度比を算出することで、前記系列データと前記所定のデータとの類似度又は相違度を算出することを特徴とする付記1に記載の情報処理装置である。
付記4に記載の情報処理装置は、前記算出手段は、前記系列データから抽出した第1特徴量と、前記所定のデータから抽出した第2特徴量との差分である差分特徴量に対して第3処理を行い、前記第3処理の処理結果に基づいて前記尤度比を算出することで、前記系列データと前記所定のデータとの類似度又は相違度を算出することを特徴とする付記1に記載の情報処理装置である。
付記5に記載の情報処理装置は、前記算出手段は、複数の前記所定データの各々に対応する複数の前記尤度比を算出し、前記判定手段は、複数の前記尤度比に基づいて、前記系列データが複数の前記所定のデータのいずれと同一対象由来であるか否かを判定することを特徴とする付記1から4のいずれか一項に記載の情報処理装置である。
付記6に記載の情報処理装置は、前記判定手段は、前記系列データと同一対象由来であると判定した1つの前記所定のデータ、又は前記系列データと同一対象由来である可能性が高いと判定した2つ以上の前記所定のデータを判定結果として出力することを特徴とする付記5に記載の情報処理装置である。
付記7に記載の情報処理装置は、複数の前記所定データに対して絞り込み処理を行い、前記判定手段の判定対象となる前記所定のデータが少なくなるように限定する限定手段を更に備えることを特徴とする付記5又は6に記載の情報処理装置である。
付記8に記載の情報処理装置は、前記系列データは、対象の画像データ及び音声データの少なくとも一方を含み、前記算出手段は、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出することを特徴とする付記1から7のいずれか一項に記載の情報処理装置である。
付記9に記載の情報処理方法は、系列データに含まれる複数の要素を取得し、前記複数の要素のうち少なくとも2つの連続する要素に基づいて、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出し、前記尤度比に基づいて、前記系列データが所定のデータと同一対象由来であるか否かを判定し、前記尤度比を算出する際に、前記系列データと前記所定のデータとの類似度又は相違度を加味することを特徴とする情報処理方法である。
付記10に記載のコンピュータプログラムは、系列データに含まれる複数の要素を取得し、前記複数の要素のうち少なくとも2つの連続する要素に基づいて、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出し、前記尤度比に基づいて、前記系列データが所定のデータと同一対象由来であるか否かを判定し、前記尤度比を算出する際に、前記系列データと前記所定のデータとの類似度又は相違度を加味するようにコンピュータを動作させることを特徴とするコンピュータプログラムである。
付記11に記載の記録媒体は、付記10に記載のコンピュータプログラムが記録されていることを特徴とする記録媒体である。
11 プロセッサ
14 記憶装置
50 データ取得部
100 尤度比算出部
110 結合部
120 比較部
130 差分演算部
140 選択部
150 操作検出部
200 判定部
300 登録データ記憶部
400 出力部
500 判定対象限定部
Claims (10)
- 系列データに含まれる複数の要素を取得する取得手段と、
前記複数の要素のうち少なくとも2つの連続する要素に基づいて、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出する算出手段と、
前記尤度比に基づいて、前記系列データが所定のデータと同一対象由来であるか否かを判定する判定手段と
を備え、
前記算出手段は、前記系列データと前記所定のデータとの類似度又は相違度を加味して、前記尤度比を算出する
ことを特徴とする情報処理装置。 - 前記算出手段は、前記系列データから抽出した第1特徴量と、前記所定のデータから抽出した第2特徴量とを結合した結合特徴量に対して第1処理を行い、前記第1処理の処理結果に基づいて前記尤度比を算出することで、前記系列データと前記所定のデータとの類似度又は相違度を加味することを特徴とする請求項1に記載の情報処理装置。
- 前記算出手段は、前記系列データから抽出した第1特徴量に対して第2処理を行い、前記第2処理後の前記第1特徴量と前記所定のデータから抽出した第2特徴量との比較結果に基づいて前記尤度比を算出することで、前記系列データと前記所定のデータとの類似度又は相違度を算出することを特徴とする請求項1に記載の情報処理装置。
- 前記算出手段は、前記系列データから抽出した第1特徴量と、前記所定のデータから抽出した第2特徴量との差分である差分特徴量に対して第3処理を行い、前記第3処理の処理結果に基づいて前記尤度比を算出することで、前記系列データと前記所定のデータとの類似度又は相違度を算出することを特徴とする請求項1に記載の情報処理装置。
- 前記算出手段は、複数の前記所定のデータの各々に対応する複数の前記尤度比を算出し、
前記判定手段は、複数の前記尤度比に基づいて、前記系列データが複数の前記所定のデータのいずれと同一対象由来であるか否かを判定する
ことを特徴とする請求項1から4のいずれか一項に記載の情報処理装置。 - 前記判定手段は、前記系列データと同一対象由来であると判定した1つの前記所定のデータ、又は前記系列データと同一対象由来である可能性が高いと判定した2つ以上の前記所定のデータを判定結果として出力することを特徴とする請求項5に記載の情報処理装置。
- 複数の前記所定データに対して絞り込み処理を行い、前記判定手段の判定対象となる前記所定のデータが少なくなるように限定する限定手段を更に備えることを特徴とする請求項5又は6に記載の情報処理装置。
- 前記系列データは、対象の画像データ及び音声データの少なくとも一方を含み、
前記算出手段は、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出する
ことを特徴とする請求項1から7のいずれか一項に記載の情報処理装置。 - 系列データに含まれる複数の要素を取得し、
前記複数の要素のうち少なくとも2つの連続する要素に基づいて、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出し、
前記尤度比に基づいて、前記系列データが所定のデータと同一対象由来であるか否かを判定し、
前記尤度比を算出する際に、前記系列データと前記所定のデータとの類似度又は相違度を加味する
ことを特徴とする情報処理方法。 - 系列データに含まれる複数の要素を取得し、
前記複数の要素のうち少なくとも2つの連続する要素に基づいて、前記系列データが所定のデータと同一対象由来であることの尤もらしさを示す尤度比を算出し、
前記尤度比に基づいて、前記系列データが所定のデータと同一対象由来であるか否かを判定し、
前記尤度比を算出する際に、前記系列データと前記所定のデータとの類似度又は相違度を加味する
ようにコンピュータを動作させることを特徴とするコンピュータプログラム。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2022572838A JPWO2022144992A1 (ja) | 2020-12-28 | 2020-12-28 | |
US18/269,681 US20240061859A1 (en) | 2020-12-28 | 2020-12-28 | Information processing apparatus, information processing method, and computer program |
PCT/JP2020/049173 WO2022144992A1 (ja) | 2020-12-28 | 2020-12-28 | 情報処理装置、情報処理方法、及びコンピュータプログラム |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2020/049173 WO2022144992A1 (ja) | 2020-12-28 | 2020-12-28 | 情報処理装置、情報処理方法、及びコンピュータプログラム |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022144992A1 true WO2022144992A1 (ja) | 2022-07-07 |
Family
ID=82259145
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2020/049173 WO2022144992A1 (ja) | 2020-12-28 | 2020-12-28 | 情報処理装置、情報処理方法、及びコンピュータプログラム |
Country Status (3)
Country | Link |
---|---|
US (1) | US20240061859A1 (ja) |
JP (1) | JPWO2022144992A1 (ja) |
WO (1) | WO2022144992A1 (ja) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009245314A (ja) * | 2008-03-31 | 2009-10-22 | Kddi Corp | 時系列データの識別装置および動画像への人物メタ情報付与装置 |
WO2018179301A1 (ja) * | 2017-03-30 | 2018-10-04 | 富士通株式会社 | 情報処理システム、情報処理装置、および情報処理方法 |
-
2020
- 2020-12-28 JP JP2022572838A patent/JPWO2022144992A1/ja active Pending
- 2020-12-28 US US18/269,681 patent/US20240061859A1/en active Pending
- 2020-12-28 WO PCT/JP2020/049173 patent/WO2022144992A1/ja active Application Filing
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009245314A (ja) * | 2008-03-31 | 2009-10-22 | Kddi Corp | 時系列データの識別装置および動画像への人物メタ情報付与装置 |
WO2018179301A1 (ja) * | 2017-03-30 | 2018-10-04 | 富士通株式会社 | 情報処理システム、情報処理装置、および情報処理方法 |
Also Published As
Publication number | Publication date |
---|---|
US20240061859A1 (en) | 2024-02-22 |
JPWO2022144992A1 (ja) | 2022-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109522843B (zh) | 一种多目标跟踪方法及装置、设备和存储介质 | |
US11610394B2 (en) | Neural network model training method and apparatus, living body detecting method and apparatus, device and storage medium | |
Tao et al. | End-to-end audiovisual speech recognition system with multitask learning | |
JP5272215B2 (ja) | ノイズ及び環境による影響を抑制することが可能な顔認識方法 | |
US9767352B2 (en) | Apparatus and method for analyzing body part association | |
Broun et al. | Automatic speechreading with application to speaker verification | |
CN109886223B (zh) | 人脸识别方法、底库录入方法、装置及电子设备 | |
US10836044B2 (en) | Robot control device and robot control method | |
JP6492746B2 (ja) | 画像処理プログラム、画像処理装置、及び画像処理方法 | |
US20210012064A1 (en) | Recording medium recording complementary program, complementary method, and information processing device | |
Ponce-López et al. | Non-verbal communication analysis in victim–offender mediations | |
WO2022144992A1 (ja) | 情報処理装置、情報処理方法、及びコンピュータプログラム | |
US11544926B2 (en) | Image processing apparatus, method of processing image, and storage medium | |
Lumini et al. | Ensemble of on-line signature matchers based on overcomplete feature generation | |
WO2020137536A1 (ja) | 人物認証装置、制御方法、及びプログラム | |
KR20160044858A (ko) | 얼굴 정보를 이용한 실종 가족 찾기 시스템 및 방법 그리고 이를 구현하는 프로그램을 기록한 기록 매체 | |
US11404064B2 (en) | Information processing apparatus and speech analysis method | |
Nijhawan et al. | Real-time Object Detection for Visually Impaired with Optimal Combination of Scores | |
US20140177950A1 (en) | Recognition device, method, and computer program product | |
US20240046704A1 (en) | Determination method and determination apparatus | |
WO2024111113A1 (ja) | 情報処理装置、情報処理方法及び記録媒体 | |
WO2022137439A1 (ja) | 情報処理システム、情報処理方法、及びコンピュータプログラム | |
WO2024079854A1 (ja) | 情報処理装置、情報処理方法、及び記録媒体 | |
JP7110293B2 (ja) | 情報処理装置、情報処理方法およびプログラム | |
JP2001126074A (ja) | パターンマッチングによるデータ検索方法及びそのプログラムを記録した記録媒体 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20968005 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2022572838 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18269681 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20968005 Country of ref document: EP Kind code of ref document: A1 |