EP2095295A1 - Procede et systeme de determination d'une probabilite de presence d'une personne dans au moins une partie d'une image et programme d'ordinateur correspondant - Google Patents
Procede et systeme de determination d'une probabilite de presence d'une personne dans au moins une partie d'une image et programme d'ordinateur correspondantInfo
- Publication number
- EP2095295A1 EP2095295A1 EP07858539A EP07858539A EP2095295A1 EP 2095295 A1 EP2095295 A1 EP 2095295A1 EP 07858539 A EP07858539 A EP 07858539A EP 07858539 A EP07858539 A EP 07858539A EP 2095295 A1 EP2095295 A1 EP 2095295A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- image
- person
- probability
- face
- detected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
- G06F18/256—Fusion techniques of classification results, e.g. of results related to same input data of results relating to different input data, e.g. multimodal recognition
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/768—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using context analysis, e.g. recognition aided by known co-occurring patterns
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/814—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level using belief theory, e.g. Dempster-Shafer
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/257—Belief theory, e.g. Dempster-Shafer
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/10—Recognition assisted with metadata
Definitions
- the present invention relates to a method for determining a probability of presence of a person in an image associated with a context. It also relates to a corresponding determination system and a computer program for this application.
- the invention relates to the field of management of images with digital contents, for example digital photos.
- the first method is the contextual description such as date of shooting, geographical position, author, keywords for image etc.
- the production of this contextual description can be automatic or manual.
- the second method is the objective description by image analysis. According to this method, the analysis of an image makes it possible to provide descriptor elements, for example of portrait, landscape, sea, mountain, etc. type. This second method also allows the recognition of people or monuments contained in a reference dictionary.
- the third method is the subjective description by textual or vocal annotation.
- the user annotates a photo so as to declare what appears to be relevant and / or what may be absent from the image, for example a relationship.
- This subjective description may also be included in a communication message of the described content.
- search engines for video images on the Internet index the images by extracting keywords from the context of the identified contents, for example the file name, the text of the hypermedia link pointing to the content or a text close enough to this link. These texts are considered a subjective description of the image. The contents can only be found from searches containing these keywords.
- the subjective description by annotation appears as a complementary mode of the objective description by image analysis.
- the voice annotation has the considerable interest to be created during the production of the content because it requires only a microphone and is very natural.
- the efficient and reliable extraction of relevant descriptor indexes remains limited today because it depends on the techniques of analysis of a speech signal, which can be noisy by the sound environment present during its recording and / or of poor quality according to the devices (micro, coding) of its recording.
- Textual annotation has the advantage of having more reliable coding (ASCII or unicode) of the information.
- ASCII or unicode many works exist in the field of text indexing. But the difficulty is today displaced in the analysis of the meaning of these textual annotations. Indeed the extraction of keywords is often insufficient to find indexed content because these keywords are often ambiguous and too general. Moreover, the extraction of key words is not as effective as for textual documents often much longer and much more redundant in information than simple textual annotations often very short.
- extracted key words can be ambiguous, as for example the first names Miguel or Nancy, which reference in the first case a person of male or female and in the second case a person or a city.
- the subject of the invention is a method for determining a probability of the presence of a person in at least a part of an image, characterized in that it comprises: a) a step of analysis of the image for determining at least one area of the image in which a person's face is detected and for associating a face detection score; for a current zone determined in step a): b) a step of analyzing the current zone by comparison of the detected face with faces of identified persons stored in the preliminarily in a database to provide a list of face identifiers of persons that may be present in the current area and a combination of an identifier detection score; and c) a step of analyzing the current area to determine gender identification scores of the detected person; d) a step of merging the face detection, identifier detection and gender identification scores to determine an area score representing the probability of presence of identified persons in the current area.
- the invention makes it possible to overcome the drawbacks of the methods of the state of the art by merging information from the various analysis steps. This fusion of information will make it possible to reduce the uncertainty and inaccuracy of the descriptor indexes relating to the persons present in at least part of the image, taking into account the probability determined in step d).
- the method comprises: a step of merging the zone scores of the zones of the image for which a person's face is detected in order to calculate a second zone score representing the probability of presence identified persons in each zone;
- the method further comprises: a step of analyzing context information associated with the image to obtain additional information on the identification of the persons mentioned in the context and association information an identity score for a given context information source;
- the context information of the image contains a voice annotation and the step of analyzing the contextual information of the image comprises a recognition step in this voice annotation of denominations of persons spoken;
- the context information of the image contains a textual annotation and the step of analyzing the context information of the image comprises a recognition step in this textual annotation of names of written persons.
- the present invention also aims at a system for determining a probability of presence of a person in at least a part of an image characterized in that it comprises: a) an image analysis module able to determine at least one area of the image in which a person's face is detected and in association a face detection score; b) a module for analyzing a current area able to compare the detected face in the current area with the faces of identified persons previously stored in a database and to provide a list of face identifiers of persons that may be present in the current field and to associate an identifier detection score; c) an analysis module of the current area able to determine gender identification scores of the person detected in the current area; and e) merging means for merging the face detection, identifier detection and gender identification scores to determine a zone score representing the probability of presence of identified persons in the current area.
- the present invention finally relates to a computer program comprising code instructions which, when this program is executed on a computer, allow the implementation of the method of determining a probability of presence of a person in at least a part of an image.
- the system of the invention also comprises: a module for analyzing a voice annotation to recognize denominations of persons spoken;
- an analysis module of a textual annotation to recognize names of written people to recognize names of written people
- a database which for a set of names of a person gives the probability that the person so named is a woman.
- the identifiers of the persons whose denomination has been recognized in a voice annotation the denominations associated with each identifier being recorded in the base Bp 1 1; the identifiers of the persons whose denomination has been recognized in a textual annotation, the denominations associated with each identifier being recorded in the database Bp 1 1.
- the merging of this information consists of identifying the compatibility or the conflict between this information.
- a descriptor index a) when the face of the same person is recognized in two different areas of the same image; or (b) where the identifier associated with a face is more likely to be attributed to a person of the opposite sex than the identified person; or (c) where the face sensor has detected only one face in an area corresponding to a sufficiently large percentage of the scanned image, and the face recognition does not have the same identifier as those detected in the annotations textual and vocal.
- the invention makes it possible to reduce: the uncertainty of the descriptor indexes that are proposed by different modules in a compatible manner; - inaccuracy by retaining only the possible descriptor indexes that are most certain;
- FIG. 1 is a block diagram illustrating the structure of a system of determining a probability of presence of a person in an image associated with a context according to the invention
- FIG. 2 is a flowchart illustrating the operation of a method for determining a probability of presence of a person in an image associated with a context according to the invention
- FIG. 3 is a flowchart illustrating the fusion of the information concerning the sex and the hypotheses on the identifier of a person identified in an image obtained by the image analysis;
- FIG. 4 is a flowchart illustrating the fusion of the information of different zones of an image where faces have been recognized obtained by the analysis of the image;
- FIGS. 5A and 5B are flowcharts illustrating the fusion of the information concerning the sex of a person identified in the image by image analysis and the assumptions on the person identifier obtained by analysis of the context of the image;
- FIG. 6 is a flowchart illustrating the final fusion of all the information obtained by analyzing the image and its context.
- FIG. 1 A system for determining the probability of presence of a person in an image associated with a context is illustrated in Figure 1.
- this system is implemented on a personal computer of a user.
- This personal computer includes means of recording digital images in the form of ".jpg”, “.gif", “.bmp” format files, etc.
- this system uses the theory of belief functions also called theory of evidence. This theory is presented in the publication of Ph. Smets and R. Kennes: "The Transferable Belief Model,” Artificial Intelligence, 66 (2): 191-234, 1994.
- This system is suitable for processing an image I, for example a digital photo, designated by the reference 1 and associated with a context 2 having a voice annotation 3 and a textual annotation 5.
- the voice annotation 3 is a sound file containing information such as the names of the people in the picture.
- Text annotation 5 is a text file containing image information such as person names.
- This system comprises various modules in order to obtain information on the persons present in the image I 1. These modules are: a module 7, noted DV, image analysis 1 providing a list of zones (Z 1 , Z 2 , ..., Z 1 , ..., Z n ⁇ of the image 1 where faces of people are detected, for each face detected by the module 7, thus corresponds to a zone Z 1 of the image 1, the set of zones forming a partition of the image 1.
- the module 7 also provides a score formed of a non-zero rational number between 0 and 1 denoted MV (DV, I, Z 1 ) for each zone Z 1 of the image I 1 where a face has been detected, corresponding to the belief that the zone Z 1 contains a face;
- a module 9, denoted RV for analyzing each zone of the list of zones of the image where faces of persons have been detected by comparison of the detected faces with faces of identified persons previously stored in a database B p 1 1 to provide a list of face identifiers of people that may be present in each image area.
- the identifiers "unknown” and “unknown”, respectively, represent male and female members, respectively, who are not part of the Lp list.
- the symbol * is used when no face is detected in the image I 1.
- the module 9 also provides for each image zone Z 1 of the image I 1, scores of between 0 and 1 rated MP ( RV, I, Z 1 ) (F 3 ), F a being a subset of ⁇ pi, p 2 , ..., p r ⁇ with each person identifier p j belonging to at least one subset F a .
- the score MP (RV, I, Z 1 ) (F a ) is an indicator of belief that the zone Z 1 of the image I 1 contains the face of a person corresponding to pi, p 2 ...
- a module 13 denoted HF, for analyzing each person's face detected in each image zone to determine the sex of the person detected for example according to a conventional method such as that described in the publication by Yi D. Cheng, Alice J. OToole, Hervé Abdi: "Classifying adults and children's faces by sex: computational investigations of subcategorical feature encoding” published in Cognitive Science 25 (2001) 819-838.
- the module 13 provides, for each zone Z 1 of the image I 1, a score between 0 and 1 denoted MHF (HF, I, Z,) (human) respectively MHF (HF, I, Z,) (female), corresponding to the probability that the zone Z 1 of the image I represents a man, respectively a woman.
- MHF HF, I, Z,
- MHF HF, I, Z,
- AV a module 15, denoted AV, for analyzing the voice annotation 3 to recognize denominations of persons spoken.
- This module provides scores between 0 and 1 rated MP (AV, I) (A 3 ), A 3 being a set of identifiers of people ⁇ p- ⁇ , p 2 ... p r ⁇ - These scores are a indicator of belief that the image I 1 contains the face of a person corresponding to either pi, p 2 ... or p r . These scores are normalized, i.e., the sum of all the scores for image I 1 is 1; and
- This module provides scores between 0 and 1 rated MP (AT, I) (C 3 ), C 3 being a set of identifiers of people ⁇ p- ⁇ , p 2 ... p r ⁇ . These scores are an indicator of belief that the image I 1 contains the face of a person corresponding to either pi, p 2 ... or p r . These scores are normalized, ie the sum of all the scores for the image I 1 is equal to 1.
- the system according to the invention also comprises a database B h / f 19, containing for a set of person names the probability that each of these denominations is attributed to a female person, which is calculated from a statistical file (not shown) recording the sex and the names associated with persons,
- a database noted
- the database B s 21 contains:
- MP (S q , I) (D 3 ) D 3 being a set of person identifiers ⁇ pi, p2- - p r ⁇ and S q being a context information source 2 of the image I 1 representing the voice annotation 3 or the textual annotation 5.
- the system according to the invention also comprises a database B ⁇ r , dex 22 containing images and a list of identifiers of persons associated with each of these images.
- the list of people identifiers for a given image is empty when creating the record of this image.
- the system according to the invention comprises a software module 23 producing from the databases 1 1, 19, 21 and 22 the probability noted P (I, P J ) that there is in the image I 1, a a person with the name p j and a piece of information marked INCERTAIN (I, p ⁇ indicating the uncertainty that one has about this probability)
- This module 23 records this information of probability and uncertainty in the base B ⁇ r , dex 22 in combination with the image I 1.
- the modules 7, 9, 13, 15 and 17, as well as the module 23, run on the user's computer or on a computer accessible thereto.
- the databases 1 1, 19, 21 and 22 are recorded in one or more volatile memories or on magnetic media accessible by this computer.
- the first part of the method consists in analyzing the image I 1 in order to extract different information on the persons present in the image I 1. This part is executed by the image analysis modules 7, 9 and 13.
- the module 7 analyzes in 24 the image I 1 to detect faces in this image I 1. It provides a list of areas where faces of people are detected and provides for each of these zones Z 1 the score of corresponding MV (DV, I, Z 1 ) face detection.
- n be the number of zones Z 1 of the image I 1 for which there exists a score MV (DV, I, Z 1 ) in the base B s 21. n thus corresponds to the number of zones Z 1 of the image I where a face was detected by the DV 7 module.
- the module 9 analyzes in 25 the n zones Z 1 where a face has been detected by the module 7 and provides a list F k of identifiers, belonging to the list L p , of persons who may be present in each image zone, as well as the corresponding score (RV, I, Z 1 ) of belief indicator of this hypothesis F k of identified persons; and the module 13 analyzes at 26 each person face detected in each image zone to determine the sex of the detected person and provides for each zone Z 1 of the image I 1 the MHF sex identification scores (HF , I, Z,) (human) and MHF (HF, I, Z,) (female).
- MHF sex identification scores HF , I, Z,
- MP (RV + DV + HF, I, Z 1 ) merging all the belief indicators MP (RV, I, Z 1 ) MHF (HF, I, Z 1 ) and MV (DV, I, Z 1 ) of the base B s 21 are computed in a process noted FusionMage 27.
- FusionMage 27 verifies that the assumptions of the module 9, concerning the presence in the zone Z 1 of persons whose identifiers are contained in a list F k , subset of L p , are compatible with the hypothesis of module 13 that this zone Z 1 contains a face of a man or a woman and according to the belief of the module 7 that the zone Z 1 contains a face.
- This process defines a new list of hypotheses, subset of L p , on the identification of the person in zone Z 1 .
- This FusionZones process 29 computes for each person identified by an identifier ft belonging to L p of the belief indicators denoted M (lmage, I, ft) or else called first probabilities P (lmage, I, ft) that this person ft is in the image I 1 and a coefficient noted INCERTAIN (lmage, I, P j ) representing the uncertainty on this belief.
- This fusion process aggregates the assumptions about the people present in each zone Z 1 and favors for each zone Z 1 the hypotheses of people present in this zone Z 1 (that is to say for which the belief indicator MP ( RV + DV + HF, I, Z 1 ) (P j ) calculated at 27 is non-zero) and not present in another zone.
- a probability P f , respectively P h , that all the zones Z 1 of the image I 1 contain only women, respectively men, is calculated from the MHF belief indicators (HF, I, Z, ) (female) respectively MHF (HF, I, Z,) (male) of the base B s 21.
- This probability P h , respectively P f is equal to the product of the scores MHF (HF, I, Z,) (female) , respectively MHF (HF, I, Z,) (human) for all the zones Z 1 of the image I 1.
- the steps 27 to 30 have thus allowed the fusion of the information resulting from the analysis of the image I 1 at 24, 25 and 26.
- the second part of the method consists in analyzing the context 2 of the image I 1 in order to extract new information on the persons present in the image I 1.
- the module analyzes the voice annotation 3 in 31 to recognize pronounced person identifiers and provides the MP scores (AV, I).
- the module 17 analyzes in 32 the textual annotation 5 to recognize written persons' identifiers and provides the scores MP (AT, I).
- the scores MP (AV, I) and MP (AT, I) provided at 31 and 32 are recorded in the base B s 21 and noted as identity scores MP (S q , I), Sq being the source of context 2 information considered (voice annotation 3 or textual annotation 5). Then, for each source S q for which it exists in the database
- the HF TestCompatibility process 33 is described in detail below with reference to FIG. 5.
- an information fusion process noted FusionAnoti designated by the reference 35 calculates a belief indicator M (Anot , I, p ' j ) from the set of sources S q noted Anot.
- This FusionAnoti 35 merger process aggregates all the assumptions of people cited in an information source by reinforcing the identifiers of people cited in different sources. It also calculates the uncertainty indicator on the belief noted as INCERTAIN (AnOt, I,
- an information fusion process noted FusionAnot2 designated by reference 37 calculates a belief indicator M (Anot, I, p ' j ) also called third probability P (Anot, I, p' j ) from the set of sources S q noted Anot.
- This FusionAnot2 merge process 37 manages the conflicts between all the assumptions of people quoted in a source by decreasing the belief of each hypothesis if different hypotheses are possible.
- FusionAnoti 35 and FusionAnot2 37 fusion processes are described in detail below.
- a final fusion process of all the information noted FusionFinal designated by the reference 39 calculates the belief indicators M (I, P J ) that the person having the identifier P j is in the image I 1 by merging the belief indicator M (Anot, I, P j ) calculated by the process FusionAnoti 35 or the process FusionAnot2 37 and M (lmage, I, P j ) calculated by the process FusionZones 29.
- the process FusionFinale 39 manages conflicts between different assumptions by decreasing the belief of each of the hypotheses if different hypotheses are possible. It also calculates an indicator Uncertainty uncertainty (I, P j ) on the belief indicators M (I, ft) and the probability P (I, P j ) that the person P j is in the image I 1.
- the non-standardized conjunctive operator will be called the fusion operator (+) ', defined as follows:
- the Merge Fusion process 27 is described with reference to the flowchart of FIG.
- the process FusionImage 27 calculates belief indicators MP (HF
- the process 27 calculates at 45 the belief indicators MP (HF, I, Z 1 ) by merging with the standard conjunctive sum operator all the indicators of belief noted
- FusionZones fusion process 29 is described with reference to the flowchart of FIG. 4. It is performed using the same standardized connective operator for managing conflicts between zones when the same person is recognized in two different zones.
- This process 29 calculates at 49, for each zone Z 1 , the probability P
- the FusionZones process 29 calculates at 51 a belief indicator denoted by MP (Z 1 ', I, Z 1 ) on the information which the zone Z 1 - can bring to the zone Z 1 in the image I 1.
- the process 29 merges into 53 with the standardized conjunctive operator the belief indicators MP (Z, ', I 1 Z 1 ) and the belief indicators MP (RV + DV + HF, I 1 Z 1 ) for any zone Z 1 'different from Z 1 to obtain the belief indicators MP (lmage, l, Z 1 ) thus,
- MP (lmage, l, Z 1 ) MP (RV + DV + HF, I, Z 1 ) (+) MP (Z 1 , Z 1 ) for each Z ( different from Z 1
- the process 29 calculates at 59 the belief indicators M (lmage, I 1 P j ) on the set of assumptions ⁇ relevant, irrelevant ⁇ representing a belief that the person identifier P j is relevant, respectively no -pertinent for the image I 1 as well:
- ALPHA-IMAGE being a constant between 0 and 1 that the overall, the image analysis process including modules 7, 9 and 13 is attributed.
- the ALPHA-IMAGE constant is set at 0.7.
- the HF TestCompatibility process 33 is described with reference to the flowchart of FIGS. 5A and 5B.
- Sq denotes a source of information of the context of the image such that there exists a set D a of person identifiers such as the belief indicator MP (S q , I) (D 3 ) is non-zero in the base B s 21;
- This set L p ' is different from L p because it is possible that the context 2 of the image I 1 refers to a person not known in the base B p 1 1, in this case, when the recognized denomination is associated with no identifier, the new identifier P j is automatically generated;
- P h denotes the probability that all zones contain a human
- - Pf is the probability that all areas contain a woman
- the probability f j is the maximum probability of the probabilities recorded in the base B h / f 19 of each denomination associated with the identifier P j in the base B p 1 1. If no denomination of P j is found in the base B h / f 19 then f j is equal to 0.5.
- the process 33 then calculates in 65 a belief indicator denoted by MP '(S q , l) by merging by a normalized conjunctive operator the belief indicators MP' (S q + HF, I) and MP '(HF, I, female )
- the process 33 then calculates at 69 a belief indicator denoted MP '(S q + HF, 1) by merging by a normalized connective operator the belief indicators MP' (S q , l) and MP '(HF, I, human )
- FusionAnoti 35 it includes the step of calculating the belief indicators M (Anot, l, p ' j ) for each identifier p' j as it exists in the information base B s 21 a set F and a source S q such that MP '(S j + HF, l,) (F)> 0. M (Anot, l, p ' j ) ⁇ relevant ⁇ and
- the process FusionAnot2 37 includes the step of calculating the belief indicators M (Anot, l, p ' j ) for each identifier p' j as it exists in the information base B s 21 a set F and a source S q such that MP '(S q + HF, 1) (F)> 0.
- the process 37 then calculates the uncertainty indicator
- UNCERTAIN (Anot, p ' j ) M (Anot, p' j ) ⁇ relevant, irrelevant ⁇
- the belief indicator M (Anot, p ' j ) M (Anot, p' j ) ⁇ relevant ⁇
- the probability P (Anot, l) (p ' j ) M (S q , l, p' j ) ⁇ relevant ⁇ .
- M (lmage, l, p ' j ) (relevant) Max (MP (RV + DV + HF, l, Z l ) ( ⁇ unknown, lnconed, * ⁇ )) for all zones Z 1 and
- M (I, p j ) M (Anot, l, p,) (+) M (lmage, l, p J ).
- the method according to the invention makes it possible to associate with each digital image created by the user and recorded in his computer at the base B ⁇ r , dex 22, the list of identifiers P j of persons having a probability P (I, P j ) greater than S pr0 b to be present in the image, the uncertainty
- the user thus has, thanks to the invention, a tool allowing him to index his images very reliably.
- the method for determining the probability and the uncertainty according to the invention achieves a very complete fusion of the information of the analysis of the image and its context in order to reduce as much as possible the limitations (noise, silence, uncertainty, imprecision) of prior art image description methods.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR0609084A FR2907243A1 (fr) | 2006-10-17 | 2006-10-17 | Procede et systeme de determination d'une probabilite de presence d'une personne dans au moins une partie d'une image et programme d'ordinateur correspondant. |
| PCT/FR2007/052109 WO2008047028A1 (fr) | 2006-10-17 | 2007-10-09 | Procede et systeme de determination d'une probabilite de presence d'une personne dans au moins une partie d'une image et programme d'ordinateur correspondant |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP2095295A1 true EP2095295A1 (fr) | 2009-09-02 |
Family
ID=37907192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP07858539A Withdrawn EP2095295A1 (fr) | 2006-10-17 | 2007-10-09 | Procede et systeme de determination d'une probabilite de presence d'une personne dans au moins une partie d'une image et programme d'ordinateur correspondant |
Country Status (3)
| Country | Link |
|---|---|
| EP (1) | EP2095295A1 (fr) |
| FR (1) | FR2907243A1 (fr) |
| WO (1) | WO2008047028A1 (fr) |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6282317B1 (en) * | 1998-12-31 | 2001-08-28 | Eastman Kodak Company | Method for automatic determination of main subjects in photographic images |
| US6611622B1 (en) * | 1999-11-23 | 2003-08-26 | Microsoft Corporation | Object recognition system and process for identifying people and objects in an image of a scene |
| EP1839213A1 (fr) | 2005-01-19 | 2007-10-03 | France Telecom | Procede de generation d'index textuel a partir d'une annotation vocale |
-
2006
- 2006-10-17 FR FR0609084A patent/FR2907243A1/fr active Pending
-
2007
- 2007-10-09 EP EP07858539A patent/EP2095295A1/fr not_active Withdrawn
- 2007-10-09 WO PCT/FR2007/052109 patent/WO2008047028A1/fr not_active Ceased
Non-Patent Citations (4)
| Title |
|---|
| "Correct System Design", vol. 3087, 1 January 2004, SPRINGER INTERNATIONAL PUBLISHING, CHAM, ISBN: 978-3-642-34221-9, ISSN: 0302-9743, article ANIL K. JAIN ET AL: "Integrating Faces, Fingerprints, and Soft Biometric Traits for User Recognition", pages: 259 - 269, XP055279957, 032548, DOI: 10.1007/978-3-540-25976-3_24 * |
| ALLINSON N M ET AL: "FACE RECOGNITION: COMBINING COGNITIVE PSYCHOLOGY AND IMAGE ENGINEERING", ELECTRONICS AND COMMUNICATION ENGINEERING JOURNAL, INSTITUTION OF ELECTRICAL ENGINEERS, LONDON, GB, vol. 4, no. 5, 1 October 1992 (1992-10-01), pages 291 - 300, XP000323712, ISSN: 0954-0695 * |
| See also references of WO2008047028A1 * |
| VICKI BRUCE ET AL: "Understanding face recognition", BRITISH JOURNAL OF PSYCHOLOGY, vol. 77, no. 3, 1 August 1986 (1986-08-01), GB, pages 305 - 327, XP055279959, ISSN: 0007-1269, DOI: 10.1111/j.2044-8295.1986.tb02199.x * |
Also Published As
| Publication number | Publication date |
|---|---|
| FR2907243A1 (fr) | 2008-04-18 |
| WO2008047028A1 (fr) | 2008-04-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102202173B (zh) | 照片自动命名方法及其装置 | |
| KR100813170B1 (ko) | 사진 의미 인덱싱 방법 및 그 시스템 | |
| EP1364316A2 (fr) | Dispositif d'extraction d'informations d'un texte a base de connaissances | |
| EP3574462A1 (fr) | Detection automatique de fraudes dans un flux de transactions de paiement par reseaux de neurones integrant des informations contextuelles | |
| Steel | Artificial intelligence and CSEM-A research agenda | |
| WO2011018867A1 (fr) | Dispositif et procédé de classement d'informations, et support d'enregistrement pouvant être lu par un ordinateur | |
| FR3068807A1 (fr) | Procede de traitement d'une image montrant un document structure comprenant une zone d'inspection visuelle a partir d'une zone de lecture automatique ou de type code-barres | |
| EP3594851A1 (fr) | Procédé de reconnaissance biométrique | |
| FR3006472A1 (fr) | Procede de recherche dans une base de donnees | |
| FR3139644A1 (fr) | Methode pour le traitement automatise d’un document pour son utilisation par des personnes en situation de handicap | |
| EP2013776A1 (fr) | Procede de de-doublonnage rapide d'un ensemble de documents ou d'un ensemble de donnees contenues dans un fichier | |
| EP2095295A1 (fr) | Procede et systeme de determination d'une probabilite de presence d'une personne dans au moins une partie d'une image et programme d'ordinateur correspondant | |
| WO2023047071A1 (fr) | Procédé de traitement d'un contenu numérique exprimant un récit d'une expérience, et applications de ce procédé pour la recommandation de contenus et/ou d'expériences | |
| CN114911971A (zh) | 融合标题信息的伪造人物视频检测方法及装置 | |
| EP1554687B1 (fr) | SystEme associatif flou de description d objets multimEdia | |
| FR2986882A1 (fr) | Procede d'identification d'un ensemble de phrases d'un document numerique, procede de generation d'un document numerique, dispositif associe | |
| EP2902927B1 (fr) | Procédé et dispositif d'étiquetage d'au moins un objet multimédia | |
| WO2018015515A1 (fr) | Procedes de partage d'opinion, equipements et programmes d'ordinateur pour la mise en oeuvre des procedes | |
| WO2006077196A1 (fr) | Procede de generation d'index textuel a partir d'une annotation vocale | |
| CN116600247A (zh) | 一种信息关联匹配方法、装置、设备以及存储介质 | |
| EP4198971A1 (fr) | Method for selecting voice contents recorded in a database, according to their veracity factor | |
| FR3130422A1 (fr) | Procédé de sélection de contenus vocaux en- registrés dans une base de données, en fonction de leur facteur de véracité. | |
| FR3139211A1 (fr) | Procédé et dispositif de rendu de contenus numériques | |
| FR3104767A1 (fr) | Procede d'aide a l'identification d'un individu recherche et produit programme d'ordinateur associe | |
| WO2024002959A1 (fr) | Procédé de classification d'images, dispositif électronique et produit programme d'ordinateur correspondant |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| 17P | Request for examination filed |
Effective date: 20090515 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC MT NL PL PT RO SE SI SK TR |
|
| DAX | Request for extension of the european patent (deleted) | ||
| 17Q | First examination report despatched |
Effective date: 20100421 |
|
| RAP1 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: ORANGE |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
| 18D | Application deemed to be withdrawn |
Effective date: 20170503 |