SYSTEM AND METHOD FOR FACIAL ANALYSIS BASED ON FACIAL AREAS OF INTEREST (ROI) IN THE VEHICLE
The system dynamically adjusts ROI selection to ensure accurate and consistent facial analysis by identifying unobstructed landmarks and using deep learning for precise feature classification, addressing issues of inconsistency and noise in conventional methods.
Patent Information
- Authority / Receiving Office
- DE · DE
- Patent Type
- Applications
- Current Assignee / Owner
- MERCEDES BENZ GROUP AG
- Filing Date
- 2024-12-17
- Publication Date
- 2026-06-18
AI Technical Summary
Conventional ROI selection methods for facial analysis in vehicles result in small, inconsistent ROIs that introduce noise and errors due to localized variations and misaligned landmarks, leading to inaccurate and unreliable analysis.
A system and method that dynamically adjusts to variations in landmark visibility by detecting unobstructed ROI landmarks, constructing predefined areas using polygon peripheries, and employing deep learning models to identify and classify facial features, ensuring accurate and consistent ROI definition.
Improves accuracy and standardization of facial analysis by minimizing noise and artifacts, providing reliable and precise feature-based analysis across varying conditions and subjects.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
TECHNICAL AREA
[0001] The present disclosure relates to the field of image processing. In particular, the present disclosure provides a system and a method for facial analysis based on facial areas of interest (ROI) in a vehicle. BACKGROUND
[0002] The current algorithm for selecting regions of interest (ROIs) for facial analysis faces the critical challenge of generating ROIs that are too small and inconsistent. The resulting ROIs often do not encompass enough spatial regions, introducing significant noise into the calculated spatial averages. This noise undermines the reliability of subsequent analyses and makes the ROI prone to errors and misinterpretations. Smaller ROIs, for example, amplify the influence of localized variations such as shadows or minor imperfections, which can misrepresent the true features of the facial region.
[0003] Conventional ROI selection methods use predefined landmarks to delineate the left cheek, right cheek, and forehead region. The process involves defining the required landmarks as arrays and creating a white-filled frame for the reference mask. The ROI is then isolated using a bitwise AND operation between the array and the mask. However, the conventional approach has two significant drawbacks: "observed ROI cropping" and "incorrect ROI cropping." Observed ROI cropping refers to cases where the ROI is cropped too narrowly, omitting important contextual information, while incorrect ROI cropping refers to the inclusion of irrelevant regions due to misaligned or faulty landmarks.Both scenarios lead to ROIs that are either too small or incorrectly located, further increasing the noise and instability in the analysis process.
[0004] Many techniques have been developed to avoid the aforementioned problems. For example, patent US8218862B2 describes an automatic design and registration of masks and feature detection for computerized skin analysis. The system and method for automatically generating and registering a mask that delineates a region of interest (ROI) within skin images are disclosed. The images can represent anatomical areas such as the face, neck, or hands, or sections such as the cheek, forehead, or nose. The mask is dynamically generated based on anatomical landmarks (e.g., eyes, nose, lips) and adapts to variations between subjects and images to ensure a standardized and reproducible ROI selection for skin analysis. The system excludes shadowed or overexposed areas to improve the accuracy of feature analysis.In addition, the system registers the skin mask under various imaging modalities (e.g. white light, UV or fluorescence) using spatial transformation techniques for a unified analysis.
[0005] Another patent document, CN112613459A, discloses a method for detecting a face-sensitive area. The method comprises the following steps: obtaining an area; capturing a face at maximum ratio; capturing a key point; expanding the key points to obtain a mask area; generating a red image of the human face; capturing a red image cheek area and a threshold thereof; and capturing a binary mask image. According to the sensitive area capture method, the forehead key points for forehead positioning are expanded and added to more comprehensively detect the sensitive area, and the red image of the face is generated for detection to improve interference resistance and increase detection accuracy.
[0006] Known systems and procedures for autonomous driving may not highlight the inherent weaknesses of the conventional approach, which lacks robustness to varying conditions such as occlusion or different facial morphologies. The reliance on static definitions and the absence of a mechanism to validate or optimize the return on interest (ROI) exacerbate the problem, leading to inaccurate bitwise operations and further worsening the issue of "false" and "observed" role cultures.
[0007] Therefore, there is a need for a system and a procedure for facial analysis based on facial interest areas (Rol) in a vehicle, which must overcome at least the problems mentioned above. SUBJECT OF THE PRESENT DISCLOSURE
[0008] A general objective of the present disclosure is to provide a system and method for facial analysis based on facial areas of interest (ROI) in a vehicle.
[0009] One objective of this disclosure is to provide a system and procedure that ensures improved accuracy and standardization by dynamically adjusting to variations in the visibility of landmarks, such as partial occlusion or incomplete detection of landmarks on the face. This eliminates variations caused by manual or fixed ROI selection methods and ensures consistent and reliable results across different images.
[0010] Another objective of the present disclosure is to provide a system and a method that improves image quality analysis and increases accuracy by excluding non-uniform skin areas.
[0011] Another objective of this disclosure is to provide a system and procedure that minimizes noise and artifacts, thereby significantly improving the reliability and precision of feature-based analysis across multiple subjects or imaging sessions. SUMMARY
[0012] The present disclosure relates to the field of image processing. In particular, the present disclosure provides a system and a method for facial analysis based on facial areas of interest (ROI) in a vehicle.
[0013] One aspect of the present disclosure relates to a facial analysis system based on facial areas of interest (ROI) in a vehicle. The system may include at least one image acquisition unit configured to capture one or more images of a user in the vehicle. The system may include a detection module coupled to the at least one image acquisition unit and configured to detect one or more landmarks corresponding to one or more facial features of the user, based on the one or more images. The system may include a control unit coupled to the at least one image acquisition unit and the detection module.The control unit can be configured to analyze the one or more detected landmarks to determine one or more unobstructed ROI landmarks and one or more obscured ROI landmarks in the one or more images. The control unit can be configured to construct a predefined area based on the one or more unobstructed ROI landmarks in the one or more images. The control unit can be configured to calculate the one or more facial features based on the one or more unobstructed ROI landmarks and the predefined area.Furthermore, the control unit can be configured to perform facial analysis of one or more calculated facial features based on the classification of one or more unobscured ROI orientation points into at least one orientation point selected from the face and one orientation point not selected from the face in the one or more images.
[0014] In some embodiments, the one or more facial features may include at least one of the following features: an eye, a jawline, an eyebrow, and a forehead hairline of the user.
[0015] In some embodiments, the control unit is configured to construct the predefined area by forming a polygon periphery based on the detection of at least three unobstructed ROI orientation points in the one or more images. Furthermore, the control unit is configured to classify orientation points selected from the face and orientation points not selected from the face based on the at least three unobstructed ROI orientation points in the one or more images.
[0016] In some embodiments, the control unit is configured to detect at least three unobstructed ROI orientation points in the one or more facial features and classify them as the orientation point selected from the face in the one or more images. The control unit can be configured to detect at least one ROI point on the user's jawline in the one or more images. The control unit can be configured to detect a predetermined number of ROI points near the user's eye in the one or more images. Furthermore, the control unit can be configured to determine at least one ROI point on the user's forehead in the one or more images based on the detection of at least one ROI point on the user's hairline and a predetermined number of ROI points near the user's eyebrows.
[0017] In some embodiments, the control unit can be configured to verify the at least three unobstructed ROI orientation points on the one or more facial features in the one or more images in order to calculate an ROI of at least one left cheek and one right cheek. The control unit can be configured to calculate the ROI of the left cheek and the right cheek in order to perform facial analysis of the one or more facial features based on the orientation point selected from the face in the one or more images.
[0018] In some embodiments, the control unit can be configured to display the reference points not selected from the face in the one or more images based on the presence of one or more unobscured ROI reference points and one or more obscured ROI reference points in the one or more images.
[0019] In some embodiments, the control unit can be configured to use a landmark detection technique based on one or more deep learning models to identify at least one of the one or more hidden ROI landmarks and the one or more unhidden ROI landmarks in the one or more images.
[0020] In some embodiments, the control unit can be configured to process the one or more images captured under at least one of the following conditions: variable lighting conditions, one or more angles, and a blurred image, in order to identify the one or more obscured ROI orientation points and the one or more unobscured ROI orientation points.
[0021] In another aspect, the present disclosure relates to a method, implemented by at least one image acquisition unit, a detection module, and a control unit, for facial analysis based on facial areas of interest (ROI) in a vehicle. The method comprises the acquisition of one or more images of a user in the vehicle by the at least one image acquisition unit. The method comprises the detection, by a detection module, of one or more landmarks corresponding to one or more facial features of the user, based on the one or more images. The method comprises the analysis of the one or more detected landmarks by a control unit to determine one or more unobstructed ROI landmarks and one or more concealed ROI landmarks in the one or more images.The method comprises the construction of a predefined area by the control unit, based on the one or more unobstructed ROI orientation points in the one or more images. The method comprises the calculation of the one or more facial features by the control unit based on the one or more unobstructed ROI orientation points and the predefined area. The method comprises the execution of the calculated facial analysis of the one or more facial features by the control unit based on the classification of the one or more unobstructed ROI orientation points into at least one orientation point selected from the face and one orientation point not selected from the face in the one or more images.
[0022] Various objects, features, aspects and advantages of the subject matter according to the invention will become clearer from the following detailed description of preferred embodiments together with the accompanying drawing figures, in which the same numbers represent the same components. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] The accompanying drawings serve to further understand the present disclosure and are an integral part of this description. The drawings illustrate exemplary embodiments of the present disclosure and, together with the description, serve to explain the principles of the present disclosure. Fig. Figure 1 shows an exemplary block diagram of a facial analysis system based on facial areas of interest (ROI) in a vehicle, according to embodiments of the present disclosure. Fig. Figure 2 shows an exemplary block diagram of a control unit according to the embodiments of the present disclosure. Fig. Figure 3 illustrates a flowchart representing a training pipeline for facial analysis based on the facial roll in the vehicle according to the embodiments of the present disclosure. Fig. Figure 4 shows a flowchart of an exemplary method for facial analysis based on facial recognition in the vehicle according to the embodiments of the present disclosure. Fig. Figure 5 shows an example of a computer system in which or with which embodiments of the system according to the embodiments of the present disclosure can be implemented. DETAILED DESCRIPTION
[0024] A detailed description of the embodiments of the disclosure illustrated in the accompanying drawings follows. The embodiments are described in sufficient detail to clearly convey the disclosure. However, the necessary level of detail is not intended to limit foreseeable variations of the embodiments; on the contrary, it is intended to cover all modifications, equivalents, and alternatives that fall within the scope of this disclosure as defined by the accompanying claims.
[0025] The embodiments described here relate to the field of image processing. In particular, the present disclosure provides a system and a method for facial analysis based on facial areas of interest (ROI) in a vehicle.
[0026] Various embodiments of the present disclosure are described with reference to the Fig. 1-5 explained in more detail. 1-5.
[0027] Fig. Figure 1 shows an exemplary block diagram of a system 102 for facial analysis based on facial areas of interest (RoI) in a vehicle 100 according to the embodiments of the present disclosure.
[0028] With reference to Fig. In this document, a system 102 for facial analysis based on facial ROI in a vehicle 100 is disclosed. The system 102 can be used in applications such as, but not limited to, facial recognition and emotion analysis, medical imaging and research, dermatological diagnosis, and the like. The system 102 can include, among other things, a control unit 104, at least one image acquisition unit 106, and a detection module 108. The at least one image acquisition unit 106 (hereinafter also referred to as "the image acquisition unit 106") can capture one or more images 302 (see Figure 1). Fig. 3) (also referred to herein as “the Images 302”) of anatomical areas of a user’s body surface, such as an eye, forehead, and the like. The Images 302 under various imaging modalities may include standard white light, ultraviolet (UV) light, polarized light, and the like.
[0029] In some embodiments, the image acquisition unit 106 can be integrated with the detection module 108 to ensure that the acquired images 302 are compatible with one or more reference points. The detection module 108 can detect one or more reference points that correspond to one or more facial features of the user based on the images 302. The one or more facial features of the user can include, among others, the eye, jawline, eyebrow, hairline, and the like.
[0030] In some embodiments, the control unit 104 can be communicatively coupled with the at least one image acquisition unit 106 and the detection module 108. The control unit 104 can be an electronic control unit configured to analyze the one or more detected landmarks in order to determine one or more unobstructed ROI landmarks and one or more obscured ROI landmarks in the images 302.
[0031] In some embodiments, the one or more reference points may refer to specific, predefined points of interest in image 302. These are typically anatomical features such as the eyes, nose, lips, eyebrows, or jawline, which can serve as reference points for identifying specific areas of image 302. The unobscured ROI reference points may refer to a subset of the detected reference points that are fully visible and not obscured by shadows, objects, hair, or other factors. One or more unobscured ROI reference points can be reliably used to define the ROI for analysis. The one or more obscured ROI reference points may refer to a subset of the detected reference points that are partially or completely obscured, making the reference points unreliable for an accurate ROI definition.One or more unobstructed ROI orientation points may be identified, but excluded from ROI calculations or adjusted through compensation procedures.
[0032] In some embodiments, the control unit 104 can be configured to construct a predefined area based on one or more unobstructed ROI orientation points in the images 302. The predefined area can involve the formation of a polygon periphery based on the detection of at least three unobstructed ROI orientation points in the images 302 to classify the orientation points selected from the face and the unobstructed ROI orientation point based on the at least three unobstructed ROI orientation points in the images 302.
[0033] In some embodiments, the control unit 104 can be configured to calculate the one or more facial features based on the one or more unobstructed ROI orientation points and the predefined area. Furthermore, the control unit 104 can be configured to detect at least three unobstructed ROI orientation points in the one or more facial features and classify the orientation point selected from the face in the images 302. The control unit 104 can be configured to detect at least one ROI point at the user's jawline in the images 302. Furthermore, the control unit 104 can be configured to detect a predetermined number of ROI points near the user's eye in the images 302.The control unit 104 can be configured to determine at least one RoI point on the user's forehead in the images 302 based on the detection of at least one RoI point on the user's hairline and a predetermined number of RoI points near the user's eyebrows.
[0034] In some embodiments, the control unit 104 can be configured to verify the at least three unobstructed ROI orientation points on the one or more facial features in Figures 302 in order to calculate an ROI of at least one left cheek and one right cheek of the user. Furthermore, the control unit 104 can be configured to calculate the ROI of the left cheek and the right cheek in order to perform facial analysis of the one or more facial features based on the orientation point selected from the face in Figures 302.
[0035] In some embodiments, the control unit 104 can be configured to perform facial analysis of one or more calculated facial features based on the classification of one or more unobscured ROI orientation points into at least one orientation point selected from the face and one orientation point not selected from the face in the images 302.
[0036] In some embodiments, the control unit 104 can be configured to display the orientation points not selected from the face in the images 302 based on the presence of one or more unobscured ROI orientation points and one or more obscured ROI orientation points in the one or more images.
[0037] In some embodiments, the control unit 104 can be configured to use a landmark detection technique based on one or more deep learning models to identify at least one of the one or more hidden ROI landmarks and the one or more unhidden ROI landmarks in the images.
[0038] In some embodiments, the control unit 104 can be configured to process the images 302 that were acquired under at least one of the following conditions: variable lighting conditions, one or more angles and a blurred image, in order to identify the one or more obscured ROI orientation points and the one or more unobscured ROI orientation points.
[0039] As in Fig. As shown in Figure 2, the system 102 can contain the components shown in block diagram 200. For example, the control unit 104 can contain one or more processors 202. The one or more processors 202 can be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, logic circuits, and / or any devices that manipulate data based on operating instructions. Among other capabilities, the one or more processors 202 can be configured to retrieve and execute computer-readable instructions stored in a memory 204 of the control unit 104. The memory 204 can hold one or more computer-readable instructions or routines that can be retrieved and executed to create or share data units via a network service.The memory 204 can include any non-volatile storage device, e.g., volatile memory such as Random-Access Memory (RAM) or non-volatile memory such as Erasable Programmable Read-Only Memory (EPROM), Flash Memory, and the like.
[0040] In one embodiment, the system 102 may also include one or more interfaces 206. The interface(s) 206 may include a variety of interfaces, such as interfaces for data input and output devices, referred to as input / output (I / O) devices, storage devices, and the like. The interface(s) 206 may enable communication between the image acquisition unit 106, the detection module 108, and the control unit 104. The interface(s) 206 may also provide a communication path for one or more components of the system 102. Examples of such components include the processing machine(s) 208 and the database 210.
[0041] In one embodiment, the processing machine(s) 208 can be implemented as a combination of hardware and programming (e.g., programmable instructions) to implement one or more functions of the processing machine(s) 208. In the examples described here, such combinations of hardware and programming can be implemented in various ways. For example, the programming for the processing machine(s) 208 can consist of processor-executable instructions stored on a non-volatile, machine-readable storage medium, and the hardware for the processing machine(s) 208 can include a processing resource (e.g., a control unit) to execute such instructions. In other embodiments, the processing machine(s) 208 can be implemented by electronic circuits.Database 210 can contain data that is either stored or generated as a result of functionalities implemented by one of the components of processing machine(s) 208.
[0042] In some embodiments, the processing machine(s) 208 may comprise an identification machine 212, a verification machine 214, a classification machine 216, and other machine(s) 218. The other machine(s) 218 may implement functions that complement the applications / functions performed by the system 102.
[0043] The identification machine 212 can be configured to detect and identify specific landmarks or features within the image 302. The identification machine 212 can locate anatomical landmarks such as the user's eyes, nose, eyebrows, jawline, and hairline, and can distinguish between obscured and unobscured landmarks to enable accurate ROI determination.
[0044] The Verification Machine 214 can be configured to validate the detected landmarks and their Returns of Interest (ROIs). The Verification Machine 214 can ensure the reliability and consistency of the identified features by checking their positions against a predefined criterion, such as landmark symmetry, orientation, or proximity. The Verification Machine 214 can eliminate false positives or poorly detected landmarks to improve accuracy.
[0045] Classification 216 can be configured to categorize identified and verified ROLS into specific predefined categories, such as the user's left cheek, right cheek, or forehead. Classification 216 ensures that ROLs are classified according to standardized anatomical regions, enabling reproducible and consistent analysis across different users and images.
[0046] Fig. Figure 3 shows a block diagram 300 representing a training pipeline for facial analysis based on facial areas of interest (ROI) according to the embodiments of the present disclosure.
[0047] In some embodiments, the system 102 can be configured to acquire the images 302 from the image acquisition unit 106. In block 304, the control unit 104 can be configured to detect one or more landmarks corresponding to the user's one or more facial features based on the images 302. The user's one or more facial features may include, but are not limited to: a left eye and a right eye, serving as the primary reference points for the upper surface; a left jaw and a right jaw, defining the lower boundary of the surface; and the like. The landmarks can be identified by a landmark detection method that accurately locates the facial features in the image 302.Furthermore, the one or more detected landmarks can be analyzed to determine the one or more unobscured ROI landmarks and the one or more obscured ROI landmarks in Figures 302.
[0048] In block 306, the control unit 104 can be configured to construct the predefined area based on one or more unobstructed ROI orientation points 304-1 in the images 302. The predefined area can be constructed by forming a polygon perimeter based on the detection of at least three unobstructed ROI orientation points 304-1 in the images 302 to classify the orientation point selected from the face and the orientation point not selected from the face in the images 302. At least 'n' unobstructed ROI orientation points are required to identify an ROI, where n ≥ 3 to form the polygon perimeter. The orientation points can define the boundaries of the ROI and ensure sufficient spatial coverage for accurate analysis.Using three or more points guarantees a stable and validated polygon for reliable ROI representation.
[0049] In block 308, the control unit 104 can check the ROI from block 306 onwards for the presence of at least one identifiable point on the jaw line and a minimum of "m" points near the eye of the area. In this way, the control unit 104 can ensure that the test ROI is accurately defined by including relevant anatomical features. Alternatively, the control unit 104 can check for the presence of "m" points near the jaw line and at least one point near the eye to further enhance the accuracy of the test ROI delineation.
[0050] In block 310, control unit 104 can define the forehead of the ROI, and control unit 104 can verify the presence of at least one identifiable point at the hairline from block 308. Additionally, in block 310-1, control unit 104 can check for the presence of detectable points near the eyebrows to ensure adequate forehead coverage. This bidirectional verification allows for robust ROI delineation and accommodates variations in facial features.
[0051] In block 312, the control unit 104 can calculate the detected ROIs for both the left cheek 312-1 and the right cheek 312-2 to determine the respective areas from block 310. By comparing the areas, the control unit 104 can identify the left cheek 312-1 and the right cheek 312-2, which may have a larger surface area.
[0052] In block 314, the control unit 104 can perform the facial analysis of one or more facial features based on the classification of one or more unobscured ROI orientation points in the orientation points not selected from the face in images 302.
[0053] Fig. Figure 4 shows a method 400 for facial analysis based on facial ROI, which can include a variety of blocks 402-412. The method 400 can be implemented by the system 102 or its control unit 104.
[0054] In Blocking 402, the procedure 400 may involve capturing one or more images 302 of a user in the vehicle 100.
[0055] In Blocking 404, the procedure 400 may include detecting one or more landmarks corresponding to one or more facial features of the user, based on one or more images 302.
[0056] In block 406, procedure 400 may include the analysis of one or more detected landmarks to determine one or more unobscured ROI landmarks and one or more obscured ROI landmarks in the one or more images.
[0057] In Block 408, Procedure 400 may involve constructing a predefined area based on the unobscured ROI orientation points in one or more images.
[0058] In block 410, procedure 400 may include the calculation of one or more facial features based on one or more unobstructed ROI orientation points and the predefined area.
[0059] In Block 412, Procedure 400 may include performing facial analysis of one or more facial features based on the classification of one or more unobscured ROI landmarks into at least one landmark selected from the face and one landmark not selected from the face in the one or more images.
[0060] System 102 and Procedure 400 can be implemented in a computer system. Fig.Figure 5 shows a block diagram of a computer system 500 comprising an external storage device 510, a bus 520, main memory 530, read-only memory 540, mass storage device 550, a communication port 560, and a processor 570. A person skilled in the art will understand that the system 500 may comprise more than one processor 570 and communication ports 560. The processor 570 may include various modules associated with the embodiments of this disclosure. The communication port 560 may be a recommended standard 232 port for use with a modem-based dial-up connection, a 10 / 100 Ethernet port, a Gigabit or 10 Gigabit port over copper or fiber optic cable, a serial port, a parallel port, or other existing or future ports. The port 560 may be selected depending on a network, such as...a Local Area Network (LAN), a Wide Area Network (WAN) or any other network to which the System 500 is connected.
[0061] In one embodiment, the memory 530 can be a RAM or other dynamic storage device generally known in the art. The read-only memory (ROM) 540 can be any static device, e.g., a programmable read-only memory (PROM) for storing static information. The mass storage 550 can be any current or future mass storage solution that can be used to store information and / or instructions. Exemplary mass storage solutions include, but are not limited to, parallel Advanced Technology Attachment (PATA) or serial Advanced Technology Attachment (SATA) hard disk drives or solid-state drives (internal or external, e.g., with Universal Serial Bus (USB) and / or FireWire interfaces), one or more optical disks, RAID (Redundant Array of Independent Disks) storage, e.g., an array of hard disks (e.g., SATA arrays).
[0062] In one embodiment, the bus 520 provides communication between the processor(s) 570 and the other storage, retention, and communication blocks. The bus 520 may be, for example, a Peripheral Component Interconnect (PCI) / PCI-Extended (PCI-X) bus, a Small Computer System Interface (SCSI), a USB bus, or similar, for connecting expansion cards, drives, and other subsystems, as well as other buses, such as a front-end bus (FSB) that connects the processor 570 to the computer system 500.
[0063] In another embodiment, operator and management interfaces, such as a display device, a keyboard, and a cursor control unit, can also be connected to the bus 520 to support direct operator interaction with the computer system 500. Other operator and management interfaces can be provided via network connections connected through the communication port 560. In some embodiments, the external storage device 510 can be any type of external hard disk drive, floppy disk drive, Compact Disc - Read Only Memory (CD-ROM), Compact Disc - Re-Writable (CD-RW), or Digital Video Disc - Read Only Memory (DVD-ROM). The components described above serve only to illustrate various possibilities. The computer system 500 mentioned is not intended to limit the scope of this disclosure in any way.
[0064] While the foregoing describes various embodiments of the present disclosure, other and further embodiments of the present disclosure may be developed without departing from the basic scope. The scope of the present disclosure is determined by the following claims. The present disclosure is not limited to the described embodiments, versions, or examples that are included to enable a person with ordinary technical knowledge to manufacture and use the present disclosure when combined with the information and knowledge available to such a person. BENEFITS OF THE PRESENT DISCLOSURE
[0065] This disclosure provides a system and procedure for facial analysis based on facial areas of interest (ROI).
[0066] This disclosure ensures improved accuracy and standardization by dynamically adjusting to variations in the visibility of landmarks, such as partial occlusion or incomplete detection of landmarks on the face. This eliminates variations caused by manual or fixed ROI selection methods and ensures consistent and reliable results across different images.
[0067] The present disclosure offers improved image quality analysis and improved accuracy by excluding inconsistent skin areas.
[0068] The present disclosure minimizes noise and artifacts, thereby significantly improving the reliability and precision of feature-based analysis across multiple subjects or imaging sessions. QUOTES INCLUDED IN THE DESCRIPTION
[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature
[0000] US 8218862B2
[0004] CN 112613459A
[0005]
Claims
[1] System (102) for facial analysis based on facial areas of interest (RoI) in a vehicle, wherein the system (102) comprises: at least one image acquisition unit (106) configured to capture one or more images (302) of a user in the vehicle; a detection module (108) coupled to the at least one image acquisition unit (106) and configured to detect one or more reference points corresponding to one or more facial features of the user based on the one or more images (302); and a control unit (104) coupled with the at least one image acquisition unit (106) and the detection module (108), wherein the control unit (104) is configured to: to analyze the one or more detected landmarks in order to determine one or more unobstructed ROI landmarks and one or more obscured ROI landmarks in the one or more images (302); to construct a predefined area based on the one or more unobscured ROI orientation points in the one or more images (302); to calculate one or more facial features based on one or more unobstructed ROI reference points and the predefined area; and to perform the facial analysis of the one or more calculated facial features based on a classification of the one or more unobscured ROI orientation points into a facially selected orientation point or a non-facially selected orientation point in the one or more images (302). [2] System (102) according to claim 1, wherein the one or more facial features comprise at least one of the following features: an eye, a jawline, an eyebrow and a forehead hairline of the user. [3] System (102) according to claim 1, wherein, for the construction of the predefined area, the processor is configured to form a polygon periphery based on the detection of at least three unobstructed ROI orientation points in the one or more images (302); classifies the orientation point selected from the face and the unobstructed ROI orientation point based on the at least three unobstructed ROI orientation points in the one or more images (302). [4] System (102) according to claim 3, wherein, for detecting at least three unobstructed ROI orientation points in the one or more facial features and for classifying the orientation points selected from the face in the one or more images (302), the control unit (104) is configured such that it: at least one roll point on the user's jaw line is detected in the one or more images (302); a predetermined number of ROI points near the user's eye are detected in the one or more images (302); and at least one RoI point on the user's forehead in the one or more images (302) is determined based on the detection of at least one RoI point on the user's hairline and a predetermined number of RoI points near the user's eyebrows. [5] System (102) according to claim 3, wherein the control unit (104) is configured such that it: the at least three unobstructed ROI orientation points on the one or more facial features in the one or more images (302) verified to calculate an ROI of at least one left cheek and one right cheek; and the RoI of the left and right cheek is calculated to enable facial analysis of one or more facial features based on the reference point selected from the face in the one or more images (302). [6] System (102) according to claim 3, wherein the control unit (104) is configured such that it: indicates the landmark not selected from the face in the one or more images (302) based on the presence of one or more unobscured ROI landmarks and the one or more obscured ROI landmarks in the one or more images (302). [7] System (102) according to claim 1, wherein the control unit (104) is configured such that it: applies a method for detecting landmarks based on one or more deep learning models to identify at least one of the one or more hidden ROI landmarks and the one or more unhidden ROI landmarks in the one or more images (302). [8] System (102) according to claim 1, wherein the control unit (104) is configured such that it: the one or more images (302) that were captured under at least one of a variable state, one or more angles and a blurred image are processed to identify the one or more obscured ROI orientation points and the one or more unobscured ROI orientation points. [9] Method for facial analysis based on facial areas of interest (ROI) in a vehicle, wherein the method comprises the following steps: Capture, by means of at least one image capture unit (106), one or more images (302) of a user in the vehicle; Detect, by means of a detection module (108), one or more landmarks corresponding to one or more facial features of the user, based on the one or more images (302); Analyzing the one or more detected landmarks by a control unit (104) to determine one or more unobscured ROI landmarks and one or more obscured ROI landmarks in the one or more images (302); Construct, by the control unit (104), a predefined area based on the one or more unobscured ROI orientation points in the one or more images (302); Calculate, by the control unit (104), the one or more facial features based on the one or more unobstructed ROI orientation points and the predefined area; and Performing facial analysis of the one or more calculated facial features by the control unit (104), based on the classification of the one or more unobscured ROI orientation points into at least one orientation point selected from the face and one unobscured orientation point in the one or more images (302).