Method and system for reconstructing the three-dimensional shape of an object from a monocular image

Abstract

Method for reconstructing the three-dimensional envelope of an object O from a monocular image 1000. The method comprises the following steps: (a) Provide 5001 a proof three-dimensional envelope E of the object O by processing the monocular image 1000; (b) Obtain 5002 a first modified three-dimensional envelope EH1 by applying, on the proof three-dimensional envelope E, a first geometric transformation; (c) Obtain 5003 a second modified three-dimensional envelope EH2 by applying, on the first modified three-dimensional envelope EH1, a second geometric transformation; (d) Obtain 5004 a third modified three-dimensional envelope EH3, by applying, on the second modified three-dimensional envelope EH2, a third geometric transformation.

Description

Title of the invention: Method and system for reconstructing the three-dimensional shape of an object from a monocular image. Technical field

[0001] The present invention relates to a method and a system for reconstructing the three-dimensional shape of an object from a monocular image. Technical background

[0002] It is common practice to detect and count individuals forming a crowd in public spaces such as streets, train stations, airports, squares, forums, pilgrimage sites, exhibition halls, concert venues, and other events. This detection and counting enables, in particular, the implementation of a number of intelligence, organizational, equipment, and logistical measures. These measures range, for example, from simple journalistic reporting for audience measurement to administrative or police measures for public safety and security, including the regulation of site attendance and the evacuation of people in the event of incidents.Statistical studies based on these actions can also provide essential information for establishing and / or optimizing fire evacuation plans, designing appropriate space layouts, or organizing traffic flow to facilitate crowd movement. They also form a framework for studying, modeling, and anticipating collective behavior during crowd movements.

[0003] Crowds exhibit highly diverse densities and spatial distributions, and are generally not homogeneous. They can, in particular, spread out around street furniture, building elements, landscape features such as trees and shrubs, or other objects, such as parked or moving vehicles. By way of example, a crowd may simply consist of a group of scattered pedestrians walking down a street, a dense group of runners or walkers during a marathon or demonstration, or even a group of essentially stationary individuals at a concert, festival, or in a train station or airport terminal.

[0004] However, beyond simply detecting and counting individuals, having a three-dimensional representation of the individuals in a crowd and the environment in which it evolves can also provide additional information on the arrangement of individuals with each other, their distribution in this environment, their interactions, and their movements. Thanks to a representation In three dimensions, it is possible to change the viewing angles to reveal details otherwise concealed by perspective effects or obstacles. For example, when examining the circumstances of a misdemeanor, crime, or offense, or simply to prevent the commission of legally reprehensible acts, a three-dimensional representation of a crowd can be a means for law enforcement to recover evidence or clues that are usually inaccessible in two-dimensional images.

[0005] However, most video surveillance systems rely on a network of monocular cameras in which each monocular camera is positioned to cover a different area of ​​the environment in which a plurality of moving elements or objects, such as individuals or vehicles, are moving. The difficulty then lies in reconstructing the three-dimensional shape of these objects from a monocular image. In particular, in the case of human individuals, this involves determining their posture and shape ("Human Pose & Shape" (HPS)).

[0006] Kocabas, Muhammed, et al. (2021), "SPEC: Seeing people in the wild with an estimated camera", Proceedings of the IEEE / CVF International Conference on Computer Vision, describes a method for estimating the pose and shape of an individual by successively applying two neural networks to a monocular image of an individual. The first network is trained to estimate the field of view and the pitch and roll angles from the image. The second neural network is trained to concatenate the camera calibration parameters estimated by the first network with the image features in order to regress the pose and shape of the individual.

[0007] Li, Zhihao, et al., (2022), "Cliff: Carrying location information in full frames into human pose and shape estimation", European Conference on Computer Vision. Cham: Springer Nature Switzerland, describes a method for estimating the pose and shape of an individual from a monocular image implementing a neural network based on an HMR architecture. The HMR part of the network takes, as input data, a cropped image of the individual, while information relating to the individual's environment is extracted from the cropped portion of the cropped image and reintroduced into the neural network immediately after encoding by the HMR part. Finally, the neural network is configured to perform a 3D projection of the joints onto the original image to allow the calculation of a 2D backprojection loss function.

[0008] Sun, Yu, et al. (2022), "Putting people in their place: Monocular regression of 3D people in depth", Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition, describes a method for estimating the pose and shape of individuals from a plurality of individuals based on a monocular image. The method is based on a neural network trained to estimate the 3D translation of all individuals by combining the image plane view with a bird's-eye view. The bird's-eye view is represented by a heatmap that estimates the probability of an individual being at a certain depth and an offset map to correct for the depth. The neural network is trained on a set of annotated images specifying the depth of the individuals. Summary of the invention Technical problem

[0009] A drawback of the available methods for estimating the position and three-dimensional shape of objects from a monocular image is their lack of flexibility. Whenever the characteristics or acquisition conditions of a monocular image differ from those of the images on which these methods were trained, these methods frequently fail to correctly estimate the position and three-dimensional shape of objects. This failure is particularly noticeable when, for example, the position of the camera that acquired the image or the variations in scale are unusual. One explanation is the scarcity, or even absence, of training sets comprising monocular images representative of all possible configurations and environmental types.

[0010] There is therefore a need for a flexible, simple and robust solution for reconstructing the three-dimensional shape of an object from a monocular image. Technical solution

[0011] According to a first aspect of the invention, a computer-implemented method is provided for reconstructing the three-dimensional envelope of an object from a monocular image. The method takes as input data a monocular image of at least one object in a scene and the position, in the real-world frame of reference, of the optical center of the camera representing the optical device for acquiring said monocular image, and provides, as output data, a three-dimensional envelope of said object in said scene. The method comprises the following steps: (a) Provide a three-dimensional proof envelope of the object by monocular image processing; (b) Obtain a first modified three-dimensional envelope by applying, on the three-dimensional test envelope, a first geometric transformation such that a representative point of a lower end of the three-dimensional test envelope is contained in a first previously defined reference plane corresponding to a ground; (c) Obtain a second modified three-dimensional envelope by applying, on the first modified three-dimensional envelope, a second geometric transformation such that a longitudinal plane representing the object is perpendicular to a second reference plane, said reference plane being defined by the position of the optical center of the camera, the representative point of a lower end and a representative point of an upper end of the first three-dimensional envelope; (d) Obtain a third modified three-dimensional envelope, by applying, on the second modified three-dimensional envelope, a third geometric transformation such that the geometric dimensions of the third modified three-dimensional envelope projected into the monocular image correspond to the original dimensions of the object in the monocular image.

[0012] Advantageous embodiments are described below.

[0013] According to a second aspect of the invention, a data processing device is provided comprising means for implementing a method according to the first aspect.

[0014] According to a third aspect of the invention, a computer program is provided comprising instructions which, when the program is executed by a computer, lead the computer to implement a method according to the first aspect.

[0015] According to a fourth aspect of the invention, a computer-readable storage medium is provided comprising instructions which, when executed by a computer, cause the computer to implement a method according to the first aspect.

[0016] According to a fifth aspect of the invention, a system is provided for reconstructing the three-dimensional shape of an object from a monocular image. Brief description of the drawings

[0017] [Fig-1] is a schematic representation of a group of individuals in a place public.

[0018] [Fig.2] is a schematic representation of a map of the three-dimensional envelopes of some individuals of the group in [Fig.1].

[0019] [Fig.3] is a schematic representation of a map of the three-dimensional envelopes of [Fig.2] from a lateral point of view.

[0020] [Fig.4] is a detailed schematic representation of group III of three-dimensional envelopes of [Fig.3].

[0021] [Fig.5] is a flowchart of a method according to the first aspect of the invention.

[0022] [Fig.6] is a schematic representation of a first geometric transformation of a three-dimensional envelope according to one embodiment.

[0023] [Fig.7] is a schematic representation of a second transformation geometric shape of a three-dimensional envelope according to a particular embodiment.

[0024] [Fig.8] is a schematic representation of the result of the second geometric transformation illustrated in [Fig.7].

[0025] [Fig.9] is a schematic representation of a third transformation geometric shape of a three-dimensional envelope according to a particular embodiment.

[0026] [Fig. 10] is a schematic representation of a three-dimensional envelope map of objects obtained from the map of [Fig.3] after application of a method in accordance with the first aspect of the invention.

[0027] [Fig. 11] is a schematic representation of a data processing device according to the second aspect of the invention. Detailed description of the implementation methods

[0028] For the purposes of the present invention, a "monocular image of at least one object" should be understood to mean an image, such as a photograph or video clipping, obtained using a monocular and depicting a scene with at least one object, such as a human being, an animal, or a vehicle. The image may represent a plurality of objects of the same or different kinds. Preferably, the image is a perspective image, that is, the objects are located at different depths of field.

[0029] With reference to [Fig.1], an example of a monocular image representing a scene containing at least one object can be a monocular image 1000 of a crowd of individuals 1002a-n circulating in a public square 1001. The image can be taken by a monocular surveillance camera or a photographic device (not shown).

[0030] Extracting a three-dimensional envelope, also called a three-dimensional model or "avatar," of an object such as an individual from a monocular image is a common operation. Within the scope of the invention, this extraction can be performed using any suitable method. Examples of extraction methods are detailed in the articles Kocabas, Muhammed, et al. (2021), "SPEC: Seeing people in the wild with an estimated camera," Proceedings of the IEEE / CVF International Conference on Computer Vision; Li, Zhihao, et al., (2022), "Cliff: Carrying location information in full frames into human pose and shape estimation," European Conference on Computer Vision. Cham: Springer Nature Switzerland; and Sun, Yu, et al. (2022), "Putting people in their place: Monocular regression of 3D people in depth," Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition.

[0031] With reference to [Fig. 2], an extraction method such as that described in Li, Zhihao, et al., (2022), "Cliff: Carrying location information in full frames into "Human pose and shape estimation," European Conference on Computer Vision. Cham: Springer Nature Switzerland, allows the extraction of a 2002a-n three-dimensional envelope for each of the individuals detected lOOla-n in the monocular image 1001 of [Fig. 1]. In Fig. 2, the set of 2002a-n three-dimensional envelopes forms a map 2000 superimposed on the original image 1001; the 2002a-n three-dimensional envelopes are superimposed on the lOOla-n individuals. The map of 2002a-n three-dimensional envelopes generally includes only the 2002a-n models; the sol 1001 is not represented.

[0032] Starting from the 2000 map of the 2002a-n three-dimensional envelopes of the lOOla-n individuals, different viewpoints of the scene in [Fig. 1] can be simulated by changing the perspective or the initial position of the camera. For example, with reference to [Fig. 3], by fictitiously changing the position of the camera (not shown) by having it follow an arc ("arc shot") from its initial position in [Fig. 2], it is possible to construct a representation of the scene from a lateral viewpoint in which the 2002a-n three-dimensional envelopes are seen from the side.

[0033] However, as explained previously and illustrated in [Fig. 3], current prior art methods fail not only to reposition the three-dimensional envelopes on the ground but also to resize them correctly according to perspective rules based on the new camera position. In [Fig. 3], the three-dimensional envelopes do not occupy a natural position. They are not aligned along an oblique direction corresponding to the ground reference plane 1001 of [Fig. 1]. On the contrary, they “float” in the air and are randomly distributed in compact groups along several oblique directions. Furthermore, the geometric dimensions of the adjacent three-dimensional envelopes are completely heterogeneous. Indeed, with reference to [Fig. 4], which shows in detail group III of [Fig. 1], the three-dimensional envelopes are not aligned along an oblique direction corresponding to the ground reference plane 1001 of [Fig. 1].3], some three-dimensional envelopes 4001, 4002, 4003 have much smaller geometric dimensions than other three-dimensional envelopes 4004, 4005, whereas they should have relatively similar dimensions for the same category of individuals (children, adults, men, women...).

[0034] With reference to [Fig. 1], 2, 3 & 5-10, according to a first aspect of the invention, a computer-implemented method 5000 is provided for reconstructing the three-dimensional envelope of an object O from a monocular image 1000. The method takes as input data a monocular image 1000 of at least one object O in a scene and the position C, in the real-world frame of reference, of the optical center of the camera 6001 representing the optical device for acquiring said image. monocular 1000, and provides, as output data, a three-dimensional envelope of said object O in said scene, the method comprises the following steps: (a) Provide 5001 a three-dimensional proof envelope E of the object O by monocular image processing 1000; (b) Obtain 5002 a first modified three-dimensional envelope EH1 by applying, on the three-dimensional test envelope E, a first geometric transformation such that a representative point PI of a lower end of the three-dimensional test envelope E is contained in a first reference plane P previously defined corresponding to a ground; (c) Obtain 5003 a second modified three-dimensional envelope EH2 by applying, on the first modified three-dimensional envelope EH1, a second geometric transformation such that a longitudinal plane representing the object is perpendicular to a second reference plane D, said reference plane D is defined by the position C of the optical center of the camera 6001, the point Pl-1 representing a lower end and a point P2-1 representing an upper end of the first three-dimensional envelope EH1; (d) Obtain 5004 a third modified three-dimensional envelope EH3, by applying, on the second modified three-dimensional envelope EH2, a third geometric transformation such that the geometric dimensions of the third modified three-dimensional envelope EH3 projected into the monocular image 1000 correspond to the original dimensions of the object O in the monocular image 1000.

[0035] A first remarkable effect of the invention is the preservation of the ground anchoring of the three-dimensional envelope EH3 of the object, regardless of the angle or viewpoint subsequently chosen to represent the scene. In particular, as illustrated in [Fig. 10], compared to [Fig. 3], when the method is applied to the objects in a crowd of objects, their three-dimensional envelopes all remain fixed and aligned along an oblique direction corresponding to the reference plane representing the ground 1001 beyond [Fig. 1].

[0036] A second remarkable effect of the invention is the preservation of the geometric proportions and orientation of the three-dimensional envelopes according to the distance and direction of observation of the camera. Thus, the posture and shape of the three-dimensional envelopes remain representative of those expected during a change of angle or viewpoint. With reference to [Fig. 10], compared to [Fig. 3], the dimensions of adjacent three-dimensional envelopes remain comparable for the same category of individuals.

[0037] The invention thus advantageously makes it possible to provide representative and realistic information on the arrangement of objects, their distribution in this environment, their interactions and their movements within the environment in which they evolve. Details otherwise concealed by perspective effects or obstacles can then be properly revealed by varying the viewing angles.

[0038] In the following discussion, for the sake of brevity, the details and various embodiments of the method according to the invention are described for a monocular image of a scene comprising a crowd of human beings. However, this approach is purely illustrative and should not be considered as limiting the invention. The method can be applied to a monocular image of a scene comprising any type of object, such as a human being, an animal, or a vehicle.

[0039] In step (a), a three-dimensional proof envelope E of the object O is obtained by processing the monocular image 1000. The processing is any type of image processing suitable for extracting a three-dimensional envelope. Preferably, the three-dimensional proof envelope (E) is obtained using a previously trained convolutional artificial neural network. By way of example, it may be a method described in any of the articles cited above, in particular in the article cited in the discussion with reference to [Fig. 2].

[0040] In step (b), with reference to [Fig. 6], a first geometric transformation is applied to the three-dimensional test envelope E such that a representative point PI of a lower end of the three-dimensional test envelope E is contained within a first previously defined reference plane P corresponding to a ground plane. In [Fig. 6], the image point of the representative point PI is denoted Pl-1 and is contained within the reference plane P. The three-dimensional envelope modified after the transformation is denoted EH1.

[0041] The representative point PI of a lower extremity of the object is of any suitable type. Its choice generally depends on the nature and position of the object. According to some preferred embodiments, when the object is a human being in an upright position, the lower extremity of the three-dimensional proof envelope E of the human being corresponds to the feet of said human being. The representative point (PI), before transformation, can then be chosen at the heel, the arch of the foot, or the toe of the individual. Alternatively, the lower extremity can be a lower limb, a lower joint, or the buttocks.In this case, since the representative point is selected in a higher area of ​​the individual's body, it is preferable to provide a compensation distance relative to the first reference plane (P) previously defined corresponding to a ground in order to prevent, at the end of the first geometric transformation, all or part of the lower extremity from ending up below said reference plane.

[0042] The first geometric transformation is of any type suitable for placing a representative point PI of a lower end of the three-dimensional proof envelope E within a first reference plane P. By way of example, it may be a translation along a direction defined relative to the viewing angle of the camera 6001, which represents the optical acquisition device of said monocular image 1000. Depending on the distance over which the three-dimensional proof envelope (E) is moved, its size may then be either smaller or larger than the expected size, from the point of view of the camera 6001, at the focal length where the three-dimensional proof envelope (E) is located after displacement. The translation operation may then be complemented by a scaling or shrinking operation so that the size of the three-dimensional envelope conforms to the expected size at that distance.

[0043] According to some preferred embodiments, the first geometric transformation is a homothety with the camera position C as the center of homothety. Such a transformation is advantageous in that it allows the representative point PI of a lower end of the three-dimensional proof envelope E to be placed in the first reference plane P while preserving the proportions.

[0044] In the example illustrated in Fig. 6, the representative point P1 of a lower end of the three-dimensional proof envelope E is displaced to an image point P1-1 along the direction C1, and the representative point P2 of an upper end of the three-dimensional proof envelope E is displaced to an image point P2-1 along the direction CP2. The algebraic ratio P1_CP2 is greater than 1; the envelope CPU “CP2-1 modified three-dimensional EH1 is an enlargement of the three-dimensional proof envelope E and its lower end is placed on the reference plane P.

[0045] In step (c), with reference to [Fig. 7] & 8, a second geometric transformation is applied to the first modified three-dimensional envelope EH1 such that a longitudinal plane representing the object is perpendicular to a second reference plane D defined by the position C of the optical center of the camera 6001, the point Pl-1 representing a lower end, and a point P2-1 representing an upper end of the second model EH1. In the example of [Fig. 7] & 8, the longitudinal plane representing the object perpendicular to the second reference plane D is defined by the two directions Z and W originating from the point Pl-1; the direction W is contained within the plane D. For a human being in an upright position, depending on their orientation, the longitudinal plane may correspond to the sagittal plane, the coronal plane, or an intermediate plane between these two planes.

[0046] Figure 7 shows the first modified envelope EH1 before the second geometric transformation. Figure 8 shows the second modified envelope EH2 after the application of the second geometric transformation on the first modified envelope EH1

[0047] The point P2-1 representing an upper extremity of the first three-dimensional envelope EH1 is of any suitable type. According to some preferred embodiments, the upper part of the first three-dimensional envelope EH1 is chosen from the head, neck, shoulders, or chest of the human individual. Preferably, the upper part corresponds to the head and the representative point P2-1 is chosen near the top of the skull of the individual represented by the first modified three-dimensional envelope (EH1).

[0048] According to some preferred embodiments, the second geometric transformation is a rotation having, as its center of rotation, the representative point Pl-1 of the modified three-dimensional envelope EH1, such that a first direction U representing a longitudinal direction of the object and a second direction W coincide. The second direction W is an orthogonal projection, onto the second reference plane D, of a direction Z orthogonal to the first reference plane P at the representative point Pl-1 of the lower part of the first modified envelope EH1. In the example of Figs. 7 & 8, the first direction U is defined by the points Pl-1 and P2-1 representing, respectively, the lower and upper parts of the modified three-dimensional envelope EH1. Alternatively, it can be defined by any other cross-section of points representing the lower and upper parts of the three-dimensional envelope EH1. In Figs.6 & 7, the rotation is represented by the angle F between the first direction U and the second direction W. The point P2-2 of the upper part of the modified three-dimensional envelope EH2 is the image of the point P2-1 representing the upper part of the modified three-dimensional envelope EH1.

[0049] In step (d), with reference to [Fig. 9], a third geometric transformation is applied to the second modified three-dimensional envelope EH2 so that the geometric dimensions of the third modified three-dimensional envelope EH3 projected onto the monocular image 1000 correspond to the original dimensions of the object O in the monocular image 1000. This step has the effect of resizing the three-dimensional envelope to compensate for any variations in its dimensions resulting from the second geometric transformation. The size of the three-dimensional envelope then conforms to that expected in the focal plane of the camera 6001 in which it is located.

[0050] In the example in [Fig. 8], after the second geometric transformation, the size of the second three-dimensional envelope is larger than it should be at this distance from the camera 6001. This difference in size is represented by the distance h between points P2-1 and P2-2. As illustrated on [Fig. 9], at the end of the third geometric transformation, the geometric dimensions of the third modified three-dimensional envelope EH3 are reduced compared to those of the second modified three-dimensional envelope EH2; the representative point P2-2 then has the representative point P2-3 as its image. The size of the three-dimensional envelope EH3 thus becomes consistent with the focal length of the camera 6001 at its distance from the camera.

[0051] According to some preferred embodiments, the third geometric transformation is a homothety with the representative point Pl-1 of the lower end of the modified three-dimensional envelope EH3 as the center of homothety. The homothety has the advantage of preserving the proportions of the three-dimensional envelope when it is resized.

[0052] In the example illustrated in Fig. 9, the representative point P2-2 of the upper end of the second three-dimensional envelope is moved to an image point P2-3 along the direction P1-1P2-2. The algebraic ratio is less than 1; the modified three-dimensional envelope EH3 is a reduction of the second modified three-dimensional envelope EH2, and the representative point P2-3 is aligned with the representative point Pl-1 of the upper end of the first three-dimensional envelope and the position C of the optical center of the camera 6001.

[0053] In step (a), during the extraction of the three-dimensional proof envelope E of the object O by processing the monocular image 1000, it is common for the image background corresponding to the scene setting to be removed. As illustrated in [Fig. 3] or [Fig. 10], for a crowd of individuals, this results in a map of three-dimensional envelopes of the individuals on a neutral background.

[0054] According to some embodiments, the method further includes a step of adding a background chosen from either a background corresponding to that of the original monocular image scene or an artificial background. Adding a background corresponding to that of the original monocular image scene allows the three-dimensional envelopes to be placed within the context of the original monocular image scene. This makes it possible to provide additional information on the potential interactions of individuals with other fixed or moving objects in the scene or on their position relative to certain specific geographical areas of the scene. Conversely, adding an artificial background, such as a neutral landscape or a simple square, can advantageously preserve the confidentiality of the scene's locations.For example, in a private setting, such as a company, preserving confidentiality can be essential to prevent the disclosure of certain know-how or trade secrets.

[0055] According to certain embodiments, the method further includes a texturing step of the modified three-dimensional envelope EH3 of the object O. This additional step can be particularly advantageous when the object is an individual and it is necessary to protect its identity by preventing the disclosure of certain biometric characteristics that would allow its identification. For example, in the context of a study, modeling, and anticipation of collective behavior during the movements of a crowd, the identity of the individuals in the crowd is generally not relevant data. It is therefore appropriate to protect this identity by texturing the three-dimensional envelopes of the individuals. Similarly, in the context of a police investigation, it can be advantageous to protect the identity of individuals in a crowd who have no relevant connection to the investigation.

[0056] According to certain embodiments, the method further comprises at least one step of changing perspective or viewing angle. Since the method according to the invention allows for a realistic representation of the arrangement of objects within the environment of a scene, a change of perspective or viewing angle can be advantageous for a better understanding of the interactions between these objects and for highlighting elements of the scene that would otherwise be hidden. In particular, where the object is a human being, such a change can reveal how they interact with their environment.

[0057] As previously mentioned, the method according to the invention can be applied to a monocular image representing a scene in which any type of object is moving. According to certain advantageous modes of use, it can be implemented for the reconstruction of a three-dimensional image of a plurality of human individuals in a public or private place. It thus allows for a study of the scene in which the distribution of individuals in space and their relationships can be precisely examined.

[0058] The method according to the first aspect of the invention is implemented by computer. With reference to [Fig. 1 1], according to a second aspect of the invention, a data processing device 11000 is provided comprising means for implementing a method according to any one of the embodiments of the first aspect of the invention.

[0059] The data processing device 11000, generally called a computer, can be used to automatically execute sequences of arithmetic or logical operations to perform tasks or actions. It can include one or more central processing units (CPUs) 11001 and / or one or more graphics processing units (GPUs) 11002, as well as a communication device 11005 between these units and the other components of the device. It can also include other electronic components such as input / output interfaces 11003, Non-volatile or volatile storage devices 11004, and communication buses for data transfer between internal components of the device or with external components. One of the input / output devices 11003 may be a connection interface for a human-machine interaction device, including a graphical user interface such as a screen for displaying human-readable information and / or sending data and / or instructions to the data processing device 11000.

[0060] According to a third aspect of the invention, a computer program is provided comprising instructions which, when the program is executed by a computer, lead the computer to implement a method according to any one of the embodiments of the first aspect of the invention.

[0061] Any type of programming language, compiled or interpreted, can be used to implement the steps of the method of the invention. The computer program may be part of a software solution, that is to say, a collection of executable instructions, code, scripts or other elements, and / or databases.

[0062] According to a fourth aspect of the invention, a computer-readable recording medium 11004 is provided comprising instructions which, when executed by a computer, cause the computer to implement a method according to any one of the embodiments of the first aspect of the invention.

[0063] The computer-readable storage medium 11004 is preferably non-volatile memory, for example, a hard drive or a solid-state drive. It may be removable storage media or non-removable storage media that is part of a computer. The computer-readable storage medium 11004 may also be volatile memory within removable media. This can facilitate the deployment of the invention at numerous production sites.

[0064] The computer-readable recording medium 11004 may be part of a computer used as a server from which executable instructions may be downloaded and, when executed by a computer, cause the computer to execute a method according to one of the embodiments described in this document.

[0065] The computer program and the storage medium 11004 on which it is recorded can be implemented in a distributed computing environment, for example, cloud computing. The instructions can be executed on a server to which one or more client computers can connect and provide encoded data as input data for a method according to any one of the embodiments of the first aspect of the invention. Once the data has been processed, the result can be downloaded and decoded on the client computer or sent directly, for example, in the form of instructions.

[0066] According to a third aspect of the invention, a system is provided for reconstructing the three-dimensional shape EH3 of an object O from a binocular image 10000. The system comprises: - an optical device for acquiring a monocular image 1000; - a data processing device 11004 according to the second aspect of the invention configured to receive a monocular image 1000 acquired by said optical acquisition device.

[0067] The optical device for acquiring a monocular image can be of any suitable type. It can be a photographic camera or a monocular camera, in particular a monocular camera for surveillance of a public or private place. In the latter case, the system can advantageously be used for the reconstruction of a three-dimensional image of a plurality of human individuals moving about in a public or private place. References Non-patent literature

[0068] Kocabas, Muhammed, et al. (2021), "SPEC: Seeing people in the wild with an estimated camera", Proceedings of the IEEE / CVF International Conference on Computer Vision.

[0069] Li, Zhihao, et al., (2022), "Cliff: Carrying location information in full frames into human pose and shape estimation", European Conférence on Computer Vision. Cham: Springer Nature Switzerland.

[0070] Sun, Yu, et al. (2022), "Putting people in their place: Monocular régression of 3D people in depth", Proceedings of the IEEE / CVF Conférence on Computer Vision and Pattern Récognition.

Claims

1. Demands Method (5000), implemented by computer, for reconstructing the three-dimensional envelope of an object (O) from a monocular image (1000), the method takes, as input data, a monocular image (1000) of at least one object (O) in a scene and the position (C), in the real-world frame of reference, of the optical center of the camera (6001) representative of the optical device for acquiring said monocular image (1000), and provides, as output data, a three-dimensional envelope of said object (O) in said scene, the method comprises the following steps: (a) Provide (5001) a three-dimensional proof envelope (E) of the object (O) by monocular image processing (1000); (b) Obtain (5002) a first modified three-dimensional envelope (EH1) by applying, on the three-dimensional test envelope (E), a first geometric transformation such that a representative point (PI) of a lower end of the three-dimensional test envelope (E) is contained in a first reference plane (P) previously defined corresponding to a ground; (c) Obtain (5003) a second modified three-dimensional envelope (EH2) by applying, on the first modified three-dimensional envelope (EH1), a second geometric transformation such that a longitudinal plane representing the object is perpendicular to a second reference plane (D), said reference plane (D) is defined by the position (C) of the optical center of the camera (6001), the point (Pl-1) representing a lower end and a point (P2-1) representing an upper end of the first three-dimensional envelope (EH1); (d) Obtain (5004) a third modified three-dimensional envelope (EH3), by applying, on the second modified three-dimensional envelope (EH2), a third geometric transformation such that the geometric dimensions of the third modified three-dimensional envelope (EH3) projected into the monocular image (1000) correspond to the original dimensions of the object (O) in the monocular image (1000).

2. Method (5000) according to claim 1, wherein first geometric transformation is a homothety with the position (C) of the camera as the center of homothety.

3. Method (5000) according to any one of claims 1 to 2, wherein second geometric transformation is a rotation having, as center of rotation, the representative point (Pl-1) of the modified three-dimensional envelope (EH1), such that a first direction (U) representative of a longitudinal direction of the object and a second direction (W) coincide, the second direction (W) is an orthogonal projection, in the second reference plane (D), of a direction (Z) orthogonal to the first reference plane (P) at the representative point (Pl-1) of the lower part of the first modified envelope (EH1).

4. Method (5000) according to any one of claims 1 to 3, wherein the third geometric transformation is a homothety with the representative point (Pl-1) of the lower end of the modified three-dimensional envelope (EH3) as the center of homothety.

5. Method (5000) according to any one of claims 1 to 4, wherein the object is a human individual in an upright position and the lower end of the three-dimensional proof envelope (E) of the human individual corresponds to the feet of said human individual.

6. Method (5000) according to claim 5, wherein the upper part of the first three-dimensional envelope (EH1) is selected from the head, neck, shoulders, or chest of the human individual.

7. Method (5000) according to any one of claims 1 to 6, wherein the three-dimensional proof envelope (E) of the object is obtained using a pre-trained convolutional artificial neural network.

8. Method (5000) according to any one of claims 1 to 7, further comprising a step of adding a background selected from a background corresponding to that of the scene of the original monocular image or an artificial background.

9. Method (5000) according to any one of claims 1 to 8, further comprising a texturing step of the modified three-dimensional envelope (EH3) of the object (O).

10. Method (5000) according to any one of claims 1 to 6, such as further comprises at least one step of changing perspective or angle of view.

11. Use of a method according to any one of claims 1 to 10 for the reconstruction of a three-dimensional image of a plurality of human individuals in a public or private place.

12. Data processing device (11000) comprising means for implementing the method according to any one of claims 1 to 10.

13. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to implement the method according to any one of claims 1 to 10.

14. Computer-readable storage medium (11004) comprising instructions which, when executed by a computer, cause the computer to implement the method according to any one of claims 1 to 10.

15. System for reconstructing the three-dimensional shape of an object (O) from a monocular image 1000, said system comprises: - an optical device for acquiring a monocular image; - a data processing device (11000) according to claim 12 configured to receive a monocular image 1000 acquired by said optical acquisition device.

Images