SYSTEM AND METHOD FOR DETECTING THE EYE VISIBILITY RANGE BASED ON REFLECTIONS ON GLASSES

Abstract

The present disclosure relates to a system 202 and a method 600 for detecting the visual field of eyes based on reflections on eyeglasses. The method 600 comprises receiving 602 image frames of a user from sensors and detecting 604 eye metrics of each eye. Furthermore, the method 600 comprises detecting 606 that the eyeglasses are being worn by the user, based on the receipt of one or more image frames, predicting 608 that reflections on the eyeglasses overlap with an area of ​​the at least one eye, and correlating 610 the eye metrics and the visual field of the at least one eye. The method 600 further comprises determining that a correlation value is below a predetermined range in order to send a warning signal to the user.Therefore, System 202 and Procedure 600 overcome the limitations of existing systems, which have problems due to hidden ocular metrics.

Description

The present disclosure relates to the field of driver assistance systems for motor vehicles. In particular, the present disclosure provides a system and a method for detecting the field of vision of the eyes based on reflections on eyeglasses, thereby improving the accuracy and reliability of eye tracking and gaze detection even under difficult conditions. Modern cars are currently equipped with driver monitoring systems (DMS) as a safety-critical component. These systems are designed to detect driver impairments such as distraction or fatigue, which are responsible for 80% of traffic accidents. Furthermore, a primary use case for artificial intelligence (AI) in DMS is the prediction of eye landmarks and driver gaze detection. The inferential performance of gaze and eye landmark prediction techniques also depends on the visibility of the driver's eyes. Most modern cars use infrared cameras (IR cameras) as the image input device for DMS modules. These IR cameras produce single-channel images, which can be processed faster compared to 3-channel RGB (red-green-blue) cameras.Furthermore, modern IR cameras are equipped with vertical-cavity surface-emitting lasers (VCSELs) that emit laser light and are commonly used in low-light scenarios, such as in vehicle interior sensors. However, if the driver is wearing glasses, these VCSELs can cause reflections off the lenses, known as VCSEL reflections. Some studies indicate that 75% of the population wears varifocal lenses, and a large percentage prefer to wear sunglasses while driving. These IR reflections impair the eyes' visibility, which in turn affects the performance of the DMS signals. Therefore, detecting VCSEL reflections on the driver's glasses is essential to ensure the reliability of the eye signals. Numerous techniques have already been developed to circumvent the aforementioned problems.For example, existing systems include circuits that receive one or more images from the left eye and one or more images from the right eye, determine the gaze direction of the left eye based on at least one received image from the left eye, determine the gaze direction of the right eye based on at least one received image from the right eye, determine a first confidence value based on the one or more received images from the left eye, determine a second confidence value based on the one or more received images from the right eye, and determine a final gaze direction at least partially based on the first and second confidence values. The first and second confidence values ​​represent indicators of the reliability of the determined gaze directions of the left eye and the right eye, respectively.Appropriate procedures and computer-readable media will also be provided. There are several other existing systems and methods for detecting a subject's gaze, including an image sensor configured to generate an image by capturing reflected light from a subject illuminated from the side by light from a light source, and a processor. The processor is configured to extract an area from the image where the intensity of the reflected light is equal to or greater than a certain value. If two or more areas result from the extraction, where a first area with a property indicating motion over time coexists with a second area with a different property indicating no motion over time, the processor identifies the first area as corneal reflection and detects the subject's gaze based on the corneal reflection. Conventional methods and systems have limitations in accurately determining gaze direction under changing conditions, as they often rely on confidence values ​​derived from individual eye images or specific areas of reflected light, potentially failing to adequately account for factors such as reflections off glasses, poor visibility, or inconsistent lighting, resulting in reduced reliability in real-world scenarios. Therefore, there is a need to at least remedy the aforementioned disadvantages and other shortcomings, or at least to provide a useful alternative to the existing procedures and systems. Aspects of the present disclosure relate to the field of an assistance system for motor vehicles. In particular, the present disclosure provides a system and a method for detecting the field of vision of the eyes based on reflections on eyeglasses, thereby improving the accuracy and reliability of eye tracking and eye recognition even under difficult conditions. One aspect of the present disclosure relates to a method for detecting an eye's field of vision based on reflections on eyeglasses. The method comprises one or more image frames of a user being received by one or more processors connected to the system from one or more sensors connected to the system, and the detection of eye metrics of each eye by the one or more processors based on the received image frames. Furthermore, the method comprises the detection by the one or more processors that the eyeglasses are being worn by the user, based on the received image frames, and the prediction by the one or more processors that one or more reflections on the eyeglasses overlap with an area of ​​at least one eye.Furthermore, the procedure includes correlating the eye metrics and the visibility range of at least one eye by the one or more processors based on the prediction and determination by the one or more processors that a correlation value is below a predetermined range in order to send a warning signal to the user. Another aspect of the present disclosure relates to a system for detecting the visual field of eyes based on reflections on eyeglasses. The system comprises one or more processors and a memory. The memory is operatively coupled to the one or more processors, and the memory comprises one or more instructions which, when executed, cause the one or more processors to receive one or more image frames of a user from one or more sensors connected to the system and to detect eye metrics of each eye based on the received image frames. Furthermore, the one or more processors are said to detect, based on the received image frames, that the eyeglasses are being worn by the user and to predict that one or more reflections on the eyeglasses overlap an area of ​​at least one eye.Furthermore, one or more processors should correlate the eye metrics and the visibility range of at least one eye based on prediction and determine that a correlation value is below a predetermined range in order to send a warning signal to the user. 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 drawings, in which the same numbers denote the same components. The accompanying drawings serve to enhance understanding of the present disclosure and are incorporated into and form 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. 1 illustrates a schematic representation of the influence of vertical-cavity surface-emitting lasers (VCSELs) on the prediction of the gaze direction and the alignment of key eye points according to the prior art. Fig. 2 illustrates a block diagram of an exemplary system for detecting a field of vision of the eyes based on reflections on spectacles according to an embodiment of the present disclosure.Figure 3 illustrates an exemplary schematic representation for the correlation of ocular metrics and a field of vision of the eyes according to an embodiment of the present disclosure. Figure 4 illustrates an exemplary schematic representation for the detection of the field of vision of the eyes based on reflections on the spectacles according to an embodiment of the present disclosure. Figure 5 illustrates an exemplary schematic representation for the prediction of an overlap between reflections on the spectacles and the field of vision of the eyes according to an embodiment of the present disclosure. Figure 6 illustrates an exemplary flowchart of a method for the detection of the field of vision of the eyes based on reflections on the spectacles according to an embodiment of the present disclosure.Figure 7 illustrates a block diagram of an exemplary computer system in which or with which embodiments of the present disclosure can be implemented. The following is a detailed description of the embodiments of the disclosure illustrated in the accompanying drawings. The embodiments are described in sufficient detail to clearly convey the disclosure. However, the level of detail is not intended to limit the expected variations of the embodiments; rather, it is intended to cover all modifications, equivalents, and alternatives that fall within the spirit and scope of the present disclosures as defined by the accompanying claims. Fig. 1 illustrates a schematic representation 100 of the influence of vertical-cavity surface-emitting lasers (VCSELs) on the prediction of the gaze direction and the alignment of the eye key points according to the prior art. Referring to Fig. 1, existing systems receive a driver image 102 in which a face is detected and glasses cause VCSEL reflections (denoted as a rectangular shaded box). The image is processed by a keypoint detection model 104 to identify the positions of the eyes. In this process, shaded patches (e.g., shaded dots) indicate keypoints with low reliability, meaning that the model is uncertain due to interference from the reflections, while unshaded patches (e.g., unshaded dots) indicate reliable keypoints where the keypoint detection model 104 is more confident.These detected key points are then used to create eye snapshots if the eyes are open, as shown in 106, and to predict a driver's gaze direction, resulting in a left- and right-eye gaze angle that indicates where the gaze direction is estimated for each eye, using a gaze detection module 108. However, in existing systems, driver fatigue can sometimes still be accurately detected even in the presence of disturbances using the left-eye closure value. Furthermore, driver distraction poses a significant problem if the right-eye gaze is incorrectly predicted due to VCSEL reflections. This incorrect prediction leads to an inaccurate overall gaze estimate, resulting in an erroneous average gaze direction and unreliable warnings. In contrast to conventional methods, the present disclosure can effectively overcome the disadvantages of existing systems by accurately detecting and taking into account VCSEL reflections on eyeglasses. Furthermore, the present disclosure employs a system capable of identifying key points and calculating the overlap between VCSEL reflections and the visibility of an eye area. By separately calculating the visibility of each eye, the system reduces the visibility of eyes affected by strong reflections, thus preventing false predictions. Moreover, the present disclosure ensures that drowsiness continues to be detected by the unaffected eye and that distractions are identified more reliably, leading to more accurate gaze detection and improved safety warnings. The embodiments described herein relate to the field of driver assistance systems for motor vehicles. In particular, this disclosure relates to a system and a method for detecting the field of vision of eyes based on reflections on eyeglasses, thereby improving the accuracy and reliability of eye tracking and gaze detection even under difficult conditions. Various embodiments relating to this disclosure are explained in detail with reference to Figures 2-7. Fig. 2 illustrates a block diagram 200 of an exemplary system 202 for detecting a visibility range of eyes based on reflections on spectacles according to an embodiment of the present disclosure. Referring to Fig. 2, the system 202 can comprise one or more processors 204, a memory 206, and an interface(s) 208. The one or more processors 204 can be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, logic circuits, and / or any devices that process data based on operating instructions. Among other things, the one or more processors 204 can be configured to retrieve and execute computer-readable instructions stored in the memory 206 of the system 202. The memory 206 can store one or more computer-readable instructions or routines that can be retrieved and executed to predict damage to a virtual vehicle. The memory 206 can comprise any non-volatile storage device, including, for example, volatile memory such as...a direct access memory (RAM), or a non-volatile memory, such as a erasable programmable read-only memory (EPROM), a flash memory, and the like. The interface(s) 208 can comprise a variety of interfaces, for example, interfaces for data input and output devices, referred to as I / O devices, storage devices, and the like. The interface(s) 208 can facilitate communication between the system 202 and various devices connected to it. The interface(s) 208 can also provide a communication path for one or more components of the system 202. Examples of such components include (but are not limited to) processing machine(s) 210, sensor module(s) 212, and a database 214. The database 214 can contain data that is either stored or generated as a result of functionalities implemented by one of the components of the processing machine(s) 210. In one embodiment, the processing machine(s) 210 can be implemented as a combination of hardware and programming (for example, programmable instructions) to implement one or more functionalities of the processing machine(s) 210. In the examples described herein, such combinations of hardware and programming can be implemented in various ways. For example, the programming for the processing machine(s) 210 can consist of processor-executable instructions stored on a non-volatile, machine-readable storage medium, and the hardware for the one or more processors 204 can include a processing resource (for example, one or more processors) to execute such instructions.In the examples presented, the machine-readable storage medium can store instructions which, when executed by the processing resource, implement the processing machine(s) 210. In such examples, the system 202 can comprise the machine-readable storage medium that stores the instructions and the processing resource for executing the instructions, or the machine-readable storage medium can be separate but accessible to both the system 202 and the processing resource. In other examples, the processing machine(s) 210 can be implemented by an electronic circuit. The processing machine(s) 210 can include an oculometric detection module 216, an overlap prediction module 218, a correlation module 220, and other modules 222. The other modules 222 can implement functionalities that complement the applications / functions executed by the processing machine(s) 210. When a driver (e.g., a user) is driving a vehicle, the system 208 can detect whether the user is distracted from driving. If the user is distracted, the system 208 can warn the user to increase safety while driving. In one embodiment, the sensor module(s) 212 can detect the user's field of vision based on reflections on the glasses and transmit image frames to the oculometric detection module 216. In another embodiment, the oculometric detection module 216 can, based on the received image frames, detect a plurality of key points around each of the user's eyes and extract the field of vision for each eye from the image frames based on the plurality of key points to determine the oculometric of each eye.In one embodiment, the ocular metrics may include (but are not limited to) the gaze of each eye, the viewing angle, the movement of each eye, the gaze direction of each eye, the pupil position of each eye, the orientation of each eye, the visual focus of each eye, the position of each eye, the gaze trajectory of each eye, the fixation of each eye, the alignment of each eye, and the like. In one embodiment, the vehicle may be (but are not limited to) a three-wheeler, a four-wheeler, a truck, a bus, and the like. In one embodiment, after the eye metrics have been acquired, the overlap prediction module 218 can detect that the user is wearing glasses based on the reception of one or more image frames. In one embodiment, the glasses can be sunglasses or any other type of eyewear (without being limited to this). In one embodiment, the overlap prediction module 218 can create a boundary frame around the glasses to detect them and generate a heat map based on the creation of this boundary frame. In one embodiment, a center point of the heat map can correspond to an area of ​​each eye, and the dimensions of the heat map can correspond to the dimensions of the glasses.In one embodiment, after the center point and dimensions of the glasses have been determined, the overlap prediction module 218 can detect the presence of reflections on the glasses and generate a subordinate heat map corresponding to each reflection on the glasses. In another embodiment, after the subordinate heat map has been generated, the overlap prediction module 218 can determine whether the dimensions of the subordinate heat map exceed a threshold. If the dimensions of the subordinate heat map exceed the threshold, the overlap prediction module 218 can, based on the generation of the subordinate heat map, form a contour and create the first subbounding frame around the contour.In one embodiment, after the first subboundary frame has been created, the overlap prediction module 218 can create a second subboundary frame around the plurality of key points and determine whether the overlap between the first and second subboundary frames exceeds a predefined threshold. If the overlap between the first and second subboundary frames exceeds the predefined threshold, the overlap prediction module 218 can predict that the reflections on the glasses overlap with the field of vision of at least one eye. In one embodiment, after predicting the overlap, the correlation module 220 can correlate the eye metrics and the visibility range of at least one eye to determine whether a confidence level (e.g., a value) of the correlation is below a predetermined range. If the value is below the predetermined range, the system 208 can send a warning signal to the user. In one embodiment, the confidence level can correspond to whether the user is concentrating on driving the vehicle without distraction. If the system 202 identifies that the user's concentration is below the predetermined range, the system 202 can warn the user by sending the warning signal to a device connected to the user. In one embodiment, the warning signal can include (but is not limited to) audible alarms, visual indicators, haptic feedback, and the like.In one embodiment, the device can be any electronic device capable of providing the warning signal (without being limited to that). Therefore, spectacle and VCSEL predictions can utilize a modified convolutional neural network (CNN) based on an hourglass architecture. This design can capture key features of image frames at various scales through repeated bottom-up and top-down processing. The CNN can use skip connections to preserve spatial information at each resolution, facilitating effective upsampling as processing progresses along the hourglass. For specific input image frames, the CNN can generate heat maps that indicate the position of an object of interest. In particular, the CNN can create separate heat maps for the left and right spectacle areas, one for each side. In contrast, VCSELs, which are semantically indistinguishable and may occur in one eye or multiple eyes, are represented by a single, unified heat map.In addition, the CNN can effectively manage the random presence and frequency of VCSELs across all eye components, resulting in a single heat map with peaks corresponding to the spectacle areas. These peaks, representing the centers of the lens boundary frames, are identified by locating the maximum values ​​in the heat map. The horizontal and vertical spread around these peaks determines the width and height of the spectacle boundary frames. To refine the boundary frames for the VCSEL heat map, a heat map value is set to 1 if it exceeds the threshold and to 0 if it falls below the threshold.Furthermore, the System 202 can use contour detection to identify the edges of areas where the values ​​are 1 and calculate the smallest possible bounding box that closely surrounds each of these detected areas, ensuring that the bounding boxes accurately capture the positions of the VCSEL reflections. This contour detection process enables a precise definition of the bounding boxes around VCSEL reflections and allows the CNN to accurately predict the presence and position of both spectacles and VCSEL reflections, even when multiple VCSELs are present in the same ocular area. Fig. 3 illustrates an exemplary schematic representation 300 for the correlation of ocular metrics and a visibility range of eyes according to an embodiment of the present disclosure. Referring to Fig. 3, a system (e.g., 202, as shown in Fig. 2) can comprise a module 300A for detecting the influence of VCSEL (vertical-cavity surface-emitting laser) reflections and a driver monitoring module 300B. In one embodiment, a sequence of the module 300A for detecting the influence of VCSEL reflections can comprise a spectacle and VCSEL detection model 302A, an overlap percentage test model 304A, and a model 306A for calculating the overall visual acuity of the eyes. In one embodiment, the spectacle and VCSEL detection model 302A can receive a driver visual field as input, which is an area of ​​the image that captures the driver's face and outputs the spectacle boundary frame and the VCSEL boundary frame. These boundary frames can be used to locate and define the areas occupied by the glasses and the VCSEL reflections on the glasses.In addition, the overlap percentage test model 304A can test the overlap percentage between the VCSEL and the eye area. Furthermore, the overlap percentage test model 304A can receive the VCSEL boundary frame and eye key points and output the VCSEL overlap percentage with the eye. In one embodiment, the model 306A can calculate the overall eye visibility by receiving the VCSEL overlap percentage as input and determining a confidence score for the overall eye visibility (e.g., a value). This value can reflect the confidence level in the correct detection of the eyes. Simultaneously, a sequence of the driver monitoring module 300B can include a keypoint detection model 302B, an eye-section extraction model 304B, and a gaze detection model 306B. In one embodiment, the keypoint detection model 302B can extract the driver's face from image frames and output the detected eye keypoints around the eyes. In another embodiment, the eye-section extraction model 304B can receive the driver's visual field and eye keypoints and extract the left eye-section and right eye-section, which are the specific areas around the eyes in the image. In another embodiment, the gaze detection module 306B can receive the left eye-section and right eye-section and output the gaze of the left eye and the gaze of the right eye.If eye visibility is low due to VCSEL reflections, the system 202 may not be able to accurately determine the gaze direction. In one embodiment, the system 202 can trigger a warning signal if the overall confidence of eye visibility (e.g., the value) is low, or if the system 202 detects unfavorable gaze behavior or eyes closed for an extended period. Therefore, to detect the influence of VCSEL reflections, module 302A provides a confirmation signal regarding the reliability of the predicted eye landmarks and gaze by predicting reflections, and reduces the confidence of the eye key points and gaze module in cases of lower visibility due to greater overlap with the eye area. Fig.Figure 4 illustrates an exemplary schematic representation 400 for the detection of a visibility range of the eyes based on reflections on a pair of glasses according to an embodiment of the present disclosure. Referring to Fig. 4, a system (e.g., 202, as shown in Fig. 2) can include a keypoint detection model 402A that can detect a multitude of keypoints on an area of ​​the eyes. Among the multitude of keypoints, the shaded points can indicate keypoints with low reliability, while unshaded points can represent reliable keypoints. Using keypoints, eye cutouts can be created if the eyes are open, as shown in 404A. The eyes can be cut out separately as the left eye and right eye, shown in 406A. After the left eye and right eye have been cut out separately, the system 202 can detect the eye metrics, shown in 408A. The system 202 can manage these keypoints with low reliability more effectively. In one embodiment, the system 202 can include an eye visibility module 402B that can identify the VCSEL boundary frame (e.g., a first subboundary frame) in which reflections (represented as a rectangular shaded field) occur, as well as the spectacle boundary frame (e.g., a boundary frame) to focus on the eye area. Furthermore, the overlap of VCSELs with the eye area 404B can be calculated, wherein the left eye has a minimal overlap (0%–5%), which is weighted down by 1 – 0.05 = 0.95, while the right eye has a significant overlap (about 80%), resulting in a significant reduction of the visibility confidence to 0.2.In addition, the visibility of the predicted eye areas is evaluated (406B), with the left eye maintaining high confidence due to minimal interference, while the confidence of the right eye is reduced due to high overlap with VCSEL reflections. Furthermore, the System 202 can calculate the overall visibility weight (408B), where the visibility weight for the left eye remains at 0.95*1 without downgrading, while the visibility weight for the right eye is reduced to 0.2*1 due to reflection interference. Using data from the left eye, the System 202 can accurately detect driver fatigue and avoid inaccurate predictions about driver distraction by, for example, relying primarily on the left eye and ignoring unreliable data from the right eye.Overall, the present disclosure can effectively mitigate the problems present in existing systems by dynamically adjusting the confidence of visual vision based on the degree of reflection interference, thus ensuring reliable predictions even when one eye is severely affected by reflections. By prioritizing the reliable eye, typically the eye with minimal reflection interference, the system 202 can improve the detection of driver drowsiness and distraction, reduce the risk of false warnings, and increase overall safety. Fig. 5 illustrates an exemplary schematic representation 500 for predicting the overlap between reflections on the glasses and a visibility area of ​​the eyes according to an embodiment of the present disclosure. Referring to Fig. 5, a VCSEL (e.g., the reflections) bounding frame and eye key points are provided as inputs, as shown in 502. The VCSEL bounding frame can define the reflection area, while the eye key points indicate the precise position of the eyes. For 8 key points ((x1, y1), (x2, y2) .... (x8, y8)), the system (e.g., 202, as shown in Fig. 2) can compute the bounding frame that encloses these points (e.g., an eye field, as shown in Fig. 5). Even though these key points may be erroneous if the eyes are not clearly visible, the predictions are generally close to the actual eye positions. Therefore, a bounding frame can enclose these distorted key points and serve as an approximate representation of the eye position.In addition, by calculating the narrowest box of key points 504, System 202 can compute the narrowest possible box or bounding frame (e.g., a second subbounding frame) that can enclose all key points around the eye, thus effectively representing the position of the eyes. Furthermore, by comparing an eye field and a VCSEL frame (e.g., a first subbounding frame), System 202 can evaluate the degree of overlap between VCSEL reflection and the eye area. This overlap is crucial for determining the ultimate influence of VCSEL reflection on eye visibility. The closer the VCSEL is to the eye, i.e., the greater the overlap with the eye area, the lower the visibility, which defines an infrared (IR) reflection influence signal.Furthermore, the system 202 in 508 can output the overlap percentage, which quantifies how much of the eye area is obscured by VCSEL reflection. Referring to Fig. 5, the overlap percentage is zero, meaning there is no overlap between the eye's visibility range and the VCSEL reflection. Even if the eye's key points are distorted due to adverse conditions and are predicted with high confidence, or if the eye's gaze is incorrectly predicted, the presence of this additional IR reflection influence signal, or visibility factor, confirms the reliability of the signals. Fig. 6 illustrates an exemplary flowchart of a method 600 for detecting a visibility range of eyes based on reflections on spectacles according to an embodiment of the present disclosure. Referring to Fig. 6, in step 602, method 600 may comprise receiving one or more image frames of a user from one or more sensors connected to a system (e.g., 202, as shown in Fig. 2). In 604, method 600 may comprise detecting ocular metrics of each eye based on receiving the one or more image frames. In 606, method 600 may comprise detecting that the spectacles are being worn by the user based on receiving the one or more image frames. In 608, method 600 may comprise predicting that one or more reflections on the spectacles overlap with the field of vision of at least one eye.In one embodiment, method 600 may comprise creating a first subboundary frame around each of the one or more reflections on the glasses and creating a second subboundary frame around the plurality of key points. In one embodiment, method 600 may comprise determining that the overlap between the first subboundary frame and the second subboundary frame exceeds a predefined threshold. In one embodiment, as a response to determining that the overlap between the first subboundary frame and the second subboundary frame exceeds the predefined threshold, method 600 may comprise determining that the one or more reflections on the glasses overlap with the field of vision of the at least one eye.In the case of 610, procedure 600 may include correlating the ocular metrics and the visual field of at least one eye based on the prediction. In the case of 612, procedure 600 may include determining that a correlation value is below a predetermined range in order to send a warning signal to the user. Fig. 7 illustrates an exemplary computer system 700 in which or with which embodiments of the present disclosure can be used. As shown in Fig. 7, the computer system 700 can comprise an external storage device 700, a bus 720, a main memory 730, a read-only memory 740, a mass storage device 750, communication ports 760, and a processor 770. A person skilled in the art will recognize that the computer system 700 can comprise more than one processor and more than one communication port. The processor 770 can comprise various modules associated with embodiments of the present disclosure.The communication interface(s) 760 can be any of the following: an RS-232 port for use with a modem-based dial-up connection, a 10 / 100 Ethernet port, a Gigabit or 10 Gigabit port using copper or fiber optic cabling, a serial port, a parallel port, or any other existing or future port(s). The communication interface(s) 760 can be selected depending on the network, such as a local area network (LAN), a wide area network (WAN), or any network to which the computer system 700 can be connected. The main memory 730 can be random-access memory (RAM) or any other dynamic storage device generally known in the art.The read-only memory 740 can be any static storage device, including but not limited to a programmable read-only memory (PROM) chip for storing static information, such as boot or BIOS (Basic Input / Output System) instructions for the 770 processor. The mass storage device 750 can be any current or future mass storage solution that can be used to store information and / or instructions. Bus 720 provides communication between the 770 processor and the other memory and communication blocks. Bus 720 can be, for example, a Peripheral Component Interconnect (PCI) / PCI Extended (PCI-X) bus, SCSI (Small Computer System Interface), Universal Serial Bus (USB), or similar for connecting expansion cards, drives, and other subsystems, as well as other buses, such as a Front Side Bus (FSB) that connects the 770 processor to the 700 computer system. Optionally, operator and management interfaces, such as a display, keyboard, and cursor control device, can also be connected to bus 720 to support direct operator interaction with computer system 700. Other operator and management interfaces can be provided via network connections linked through communication port(s) 760. The exemplary computer system 700 described above is not intended to limit the scope of this disclosure in any way. While the foregoing describes various embodiments of the disclosure, other and further embodiments of the invention can be developed without deviating from its fundamental scope. The scope of the disclosure is determined by the subsequent claims. The disclosure is not limited to the described embodiments, versions, or examples, which are included to enable a person with average technical knowledge to produce and use the disclosure in combination with the information and knowledge available to a person with average technical knowledge.

Claims

Method (600) for detecting an eye visibility area based on reflections on spectacles, comprising: Receiving (602) by one or more processors (204) connected to a system (202), one or more image frames of a user from one or more sensors connected to the system (202); Detecting (604) eye metrics of each eye by the one or more processors (204) based on the receipt of the one or more image frames; Detecting (606) by the one or more processors (204) that the spectacles are being worn by the user, based on the receipt of the one or more image frames; Predicting (608) by the one or more processors (204) that one or more reflections on the spectacles overlap an area of ​​at least one eye;Correlating (610) the ocular metrics and the visual field of at least one eye by the one or more processors (204) based on the prediction; and determining (612) by the one or more processors (204) that a value of the correlation is below a predetermined range in order to send a warning signal to the user. Method (600) according to claim 1, wherein the eye metrics comprise at least one of the following elements: a gaze of each eye, a gaze angle, a movement of each eye, a gaze direction of each eye, a pupil position of each eye, an orientation of each eye, a visual focus of each eye, a position of each eye, a gaze trajectory of each eye, a fixation of each eye and an alignment of each eye. Method (600) according to claim 2, wherein the recognition (604) of the eye metrics by the one or more processors (204) comprises: recognizing a plurality of key points around each eye of the user by the one or more processors (204) based on receiving the one or more image frames; and extracting the area of ​​each eye from the one or more image frames by the one or more processors (204) based on the plurality of key points for recognizing the eye metrics. Method (600) according to claim 1, wherein the detection (606) by the one or more processors (204) that the glasses are being worn by the user comprises: creating a bounding frame around the glasses in the one or more image frames by the one or more processors (204); and generating a thermal map by the one or more processors (204) based on the creation of the bounding frame. Method (600) according to claim 4, wherein a center point of the heat map corresponds to an area of ​​each eye and wherein the dimensions of the heat map correspond to the dimensions of the glasses. Method (600) according to claim 3, wherein the prediction (608) by the one or more processors (204) that the one or more reflections on the glasses overlap with the area of ​​the at least one eye comprises: creating a first subboundary frame around each of the one or more reflections on the glasses by the one or more processors (204); creating a second subboundary frame around the plurality of key points by the one or more processors (204); determining by the one or more processors (204) that the overlap between the first subboundary frame and the second subboundary frame exceeds a predefined threshold;and in response to the determination that the overlap between the first sublimiting frame and the second sublimiting frame exceeds the predefined threshold, determining by the one or more processors (204) that the one or more reflections on the spectacle overlap with the area of ​​the at least one eye. The method (600) according to claim 6, wherein the creation of the first subboundary frame by the one or more processors (204) comprises: detecting the presence of the one or more reflections on the glasses by the one or more processors (204); generating a sub-heat map by the one or more processors (204) corresponding to each of the one or more reflections on the glasses; determining by the one or more processors (204) that the dimensions of the sub-heat map exceed a threshold; and in response to the determination that the dimensions of the sub-heat map exceed the threshold, generating the first subboundary frame around the sub-heat map by the one or more processors (204). Method (600) according to claim 7, comprising: forming a contour by the one or more processors (204) based on the creation of the subordinate thermal map; and creating a first subboundary frame around the contour by the one or more processors (204). System (202) for detecting an eye's visibility area based on reflections on spectacles, comprising: one or more processors (204); and a memory (206) operatively coupled to the one or more processors (204), the memory (206) comprising one or more instructions which, when executed, cause the one or more processors (204) to: receive one or more image frames of a user from one or more sensors connected to the system (202); detect eye metrics of each eye based on the receipt of the one or more image frames; detect that the spectacles are being worn by the user based on the receipt of the one or more image frames; predict that one or more reflections on the spectacles overlap an area of ​​at least one eye;Correlating the eye metrics and the visual field of at least one eye based on the prediction; and determining when a correlation value falls below a predetermined range to send a warning signal to the user. System (202) according to claim 9, wherein the one or more processors (204) are to perform the following: creating a first subboundary frame around each of the one or more reflections on the glasses; creating a second subboundary frame around the plurality of key points; determining that the overlap between the first subboundary frame and the second subboundary frame exceeds a predefined threshold; and in response to the determination that the overlap between the first subboundary frame and the second subboundary frame exceeds the predefined threshold, determining that the one or more reflections on the glasses overlap with the area of ​​the at least one eye.

Images