Object detection device, object detection method, and recording medium having a program recorded

Abstract

An object detection device detects a given object from an image, characterized by having: a first detection unit that detects a candidate region in which the object is present from the image; a determination unit that decides an object region from one or more of the candidate regions detected by the first detection unit; a second detection unit that detects the object by a detection algorithm different from that of the first detection unit, with the object region as a target; and a storage unit that stores detection information indicating a detection result based on the second detection unit for the object region, the determination unit deciding the object region from one or more of the candidate regions based on the detection information for one or more previous frames.

Description

TECHNICAL FIELD

[0001] The present application relates to a technology of detecting an object. BACKGROUND

[0002] In the past, in object detection, a technology of performing detection using a detector divided into two stages of a front stage and a rear stage is known. For example, in Patent Literature 1, Patent Literature 2, a candidate region of a detection target (for example, a face) is detected by a detector of the front stage, and the detection target is detected from the plurality of candidate regions by a detector of the rear stage, thereby performing high-precision object detection.

[0003] PRIOR ART DOCUMENTS

[0004] PATENT LITERATURE

[0005] Patent Literature 1: Japanese Patent Application Publication No. 2006-293720

[0006] Patent Literature 2: Japanese Patent Application Publication No. 2019-021001 SUMMARY

[0007] PROBLEMS TO BE SOLVED BY THE INVENTION

[0008] However, in the related art, since two-stage detection processing is performed, there is a problem that the processing time increases. In particular, in a fixed camera, in addition to an object of a detection target (for example, a moving body), an appearance does not change, and thus a false detection once generated in the detector of the front stage is generated multiple times at the same place. For the region that is false detected, detection processing is also performed again by the detector of the rear stage, and there is a problem that the processing time further increases.

[0009] An object of the present application is to provide a technology capable of performing object detection at high speed and with high precision.

[0010] MEANS FOR SOLVING THE PROBLEMS

[0011] In order to achieve the above object, the present application adopts the following structure.

[0012] A first aspect of the present application is an object detection device that detects a given object from an image, the object detection device characterized by having: a first detection unit that detects a candidate region in which the object is present from the image; a determination unit that decides an object region from one or more of the candidate regions detected by the first detection unit; a second detection unit that detects the object by a detection algorithm different from that of the first detection unit, with the object region as a target; and a storage unit that stores detection information indicating a detection result based on the second detection unit for the object region, the determination unit deciding the object region from one or more of the candidate regions based on the detection information for one or more previous frames.

[0013] The object to be detected is not particularly limited, and examples include a human body, a face, a specific animal, a car, a specific product, and the like. The candidate region is a region in which the first detection unit determines that the probability of presence of the object to be detected is high, and the region (object region) that is the target of detection by the second detection unit is decided based on this candidate region. The first detection unit and the second detection unit can use any algorithm, but it is preferable that the detection algorithm of the second detection unit be able to detect with higher precision and have a larger amount of calculation than the detection algorithm of the first detection unit. The detection information is information obtained by performing the object detection processing by the second detection unit, and includes, for example, the position and size of the object region, an image corresponding to the object region, a score indicating the degree of certainty that the object to be detected is included in the object region, and the like.

[0014] The detection information can also include information related to an object region in which the object is not detected by the second detection unit. In this case, the determination unit can also decide, as the object region, a region in the candidate region that is similar to the object region in which the object is not detected in the previous frame to a degree of a given value or more, other than the candidate region. Furthermore, the first detection unit can also output a first detection reliability indicating the degree of certainty that the object is included in the candidate region, and the determination unit can also decide the object region based on a value obtained by subtracting a given value from the first detection reliability, for the candidate region that is similar to the object region in which the object is not detected in the previous frame to a degree of a given value or more, and decide the object region based on the first detection reliability, for the other candidate regions. According to the above-described structure, the number of candidate regions submitted to the second detection unit is reduced, and thus by performing two-stage detection processing, it is possible to reduce the processing time while maintaining the detection performance.

[0015] Further, the given value to be subtracted from the first detection reliability can also be a value corresponding to the number of consecutive frames in which the object is not detected by the second detection unit. For example, the given value can be made larger as the number of consecutive frames increases, or the given value to be subtracted from the first detection reliability can be subtracted for the first time when the number of consecutive frames is a certain number or more. Alternatively, the given value to be subtracted from the first detection reliability can be a fixed value.

[0016] Further, the first detection unit can also output a first detection reliability indicating the degree of certainty that the object is included in the candidate region, and the detection information can also include a second detection reliability determined by the second detection unit indicating the degree of certainty that the object is included in the target region. The determination unit can determine the target region based on a value obtained by subtracting a value corresponding to the second detection reliability from the first detection reliability, for the candidate region having a degree of similarity to the target region indicated by the detection information that is a given value or more, and determine the target region based on the first detection reliability, for the other candidate regions. For example, the higher the second detection reliability, the larger the given value to be subtracted from the first detection reliability can be.

[0017] The detection information can also include the position and / or size of the target region, and the determination unit can calculate the degree of similarity based on the position and / or size of the candidate region and the position and / or size of the target region. In object detection, false detection of the same object in the input image can occur multiple times, but according to the above-described configuration, multiple false detections of the same object at the same position and size can be effectively reduced. Thus, the number of candidate regions submitted to the second detection unit is reduced, and therefore, by performing two-stage detection processing, the processing time can be reduced while maintaining the detection performance.

[0018] The detection information can also include an image corresponding to the target region, and the determination unit can calculate the degree of similarity based on the image included in the detection information and the image corresponding to the candidate region. Thus, in a case where the position and size of the region corresponding to the false detection information and the candidate region are identical or similar, but the images corresponding to the two regions are completely different, object detection can be performed with high accuracy.

[0019] The second aspect of the present application is an object detection method of detecting a given object from an image, the object detection method characterized by having: a first detection step of detecting a candidate region in which the object is present from the image; a determination step of deciding an object region from one or more of the candidate regions detected in the first detection step; a second detection step of detecting the object by a detection algorithm different from that of the first detection step, with the object region as a target; and a storage step of storing detection information indicating a detection result in the second detection step with respect to the object region, in the determination step, the object region is decided from one or more of the candidate regions based on the detection information with respect to one or more previous frames.

[0020] The present application can be understood as an object detection device having at least a part of the above-described units, as a device that recognizes or tracks an object of detection, or as an image processing device or a monitoring system. Furthermore, the present application can be understood as an object detection method, an object recognition method, an object tracking method, an image processing method, or a monitoring method including at least a part of the above-described processes. Furthermore, the present application can be understood as a program for implementing the method, or a recording medium that non-transitorily records the program. In addition, each of the above-described units and processes can be combined with each other as much as possible to constitute the present application.

[0021] Effects of Invention

[0022] According to the present application, object detection can be performed at high speed and with high accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a diagram showing an application example of object detection.

[0024] Figure 2 is a diagram showing the structure of an object detection device.

[0025] Figure 3 is a flowchart of an object detection process.

[0026] Figure 4 is a flowchart of a determination process.

[0027] Figure 5 is a flowchart of a determination process. DETAILED DESCRIPTION

[0028] (Application Example)

[0029] Reference Signs Figure 1application example of the object detection device according to the present application. The object detection device detects an object (e.g., a human body) from an image obtained by a fixed camera installed above a detection target region (e.g., a ceiling). Further, the object detection device uses a detector having a two-stage structure of a front stage and a rear stage. The object 101 and the object 102 are moving bodies that move within the imaging range of the fixed camera 1 and are detection targets (e.g., human bodies). The object 103 is an object (e.g., a flower) disposed within the imaging range of the fixed camera 1. The object detection device detects candidate regions 111 to 113 in which the object exists using the detector of the front stage described above with respect to an input image. The candidate regions 111 to 113 are regions corresponding to the objects 101 to 103. The object 103 is not a human body that is a detection target, but the candidate region 113 is generated in a case where the feature of the object 103 is similar to that of a human body. Then, the object detection device performs object detection using the detector of the rear stage described above and records the detection result in a storage device. The detector of the rear stage basically performs detection with respect to object regions 121 to 123 corresponding to the candidate regions 111 to 113. Here, the detector of the front stage erroneously detects the object (flower) 103 as an object, but the detector of the rear stage can detect that it is not an object. In this case, it is considered that the detector of the front stage continuously erroneously detects the object 103. If all of the candidate regions are object regions of the detector of the rear stage, the detector of the rear stage performs detection processing every frame even in a case where there is no object, resulting in useless processing. Figure 1

[0030] Therefore, in the present application example, from the regions in which the detector of the front stage detects an object (candidate regions), the region in which the detector of the rear stage performs object detection (object region) is decided based on detection information with respect to one or more previous frames. For example, it is considered that a candidate region of the current frame that is similar to a region in which the detector of the rear stage does not detect an object in one or more previous frames is excluded from the object region. Alternatively, the object region can be decided based on the detection score (reliability) of the detector of the front stage among the candidate regions, but for a region in which the detector of the rear stage does not detect an object in one or more previous frames, the region is decided as an object region based on a value obtained by subtracting a given value from the detection score. The value to be subtracted can be a fixed value or can be set to a value corresponding to the number of continuous frames in which an object is not detected. In this way, even for a region in which the detector of the front stage detects an object, the region is excluded from the processing target of the detector of the rear stage in a case where the region is similar to a region in which the detector of the rear stage does not detect an object, and thus it is possible to maintain the accuracy of object detection and speed up the processing.

[0031] (Embodiment 1)

[0032] <Structure>

[0033] Figure 2 ​Fig. 1 is a functional block diagram of an object detection device 10 according to the present embodiment. The object detection device 10 is an information processing device (computer) including an arithmetic device (CPU; processor), a memory, a storage device (storage section 16), an input / output device, and the like. The object detection device 10 executes a program stored in the storage device, thereby providing the functions of an image input section 11, a first detection section 12, a determination section 13, a second detection section 14, an output section 15, and the like. Part or all of these functions can also be implemented by a dedicated logic circuit such as an ASIC or an FPGA.

[0034] The image input section 11 has a function of taking in image data from the camera 20. The taken-in image data is submitted to the first detection section 12. The image data can also be stored in the storage section 16. In the present embodiment, the image data is received directly from the camera 20, but can also be received via a communication device or the like, or via a recording medium. The input image is not particularly limited and can be an RGB image, a gray-scale image, an image representing a distance or a temperature, or the like.

[0035] The first detection section 12 detects a candidate region (a region in which an object of a detection target can exist) from the input image. In the present embodiment, the first detection section 12 performs detection of a candidate region using a detector that utilizes a Haar-like feature amount and adaboost. The detection result is submitted to the determination section 13. The detection result can include a detected candidate region, and further can include a degree of certainty that an object of a detection target exists in the candidate region (a first detection reliability, a detection score). The feature amount used for detection and the learning algorithm of the detector are not particularly limited. For example, as the feature amount, any feature amount such as a HoG (Histogram of Gradient) feature amount, a SIFT feature amount, a SURF feature amount, a Sparse feature amount, or the like can be used. Further, the learning algorithm can use any learning method other than adaboost, such as a boosting method, a SVM (Support Vector Machine), a neural network, a decision tree learning, or the like.

[0036] The determination section 13 determines a region (target region) that is a detection target of the second detection section 14 from the candidate regions detected by the first detection section 12. In the present embodiment, the determination section 13 determines the target region from the candidate regions using the detection information of the preceding frame stored in the storage section 16. The detection information includes information about a target region (false detection region) in which an object is not detected by the second detection section 14 described later in one or more preceding frames. The determination section 13 determines a region other than a candidate region having a degree of similarity to the false detection region of a given value or more as the target region, and outputs it to the second detection section 14 in the subsequent stage. In addition, in a case where the first detection reliability described above is included in the detection result of the first detection section 12, the determination section 13 can also determine a region from which a candidate region similar to the false detection region is removed from a candidate region having a first detection reliability of a given value or more as the target region.

[0037] The second detection section 14 performs object detection on the target region determined by the determination section 13. Information indicating whether or not an object that is a detection target exists in the target region can also be included in the detection result, and further, a degree of certainty (second detection reliability, detection score) that an object that is a detection target exists in the target region, and the like can also be included. In addition, in the present embodiment, the second detection section 14 records, as detection information, a result of the object detection, a position and / or a size of the target region in which it is determined that an object that is a detection target does not exist, in the storage section 16. In addition, the second detection section 14 can also record detection information (position and / or size) of all of the target regions determined by the determination section 13 in the storage section 16. In the present embodiment, the second detection section 14 performs detection of an object using a detector that utilizes deep learning. In addition, the method of deep learning is not particularly limited, and for example, it can also be a detector based on any method of CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), SAE (Stacked Auto Encoder), DBN (Deep Belief Network), and the like. In addition, the second detection section 14 can also not be a detector that utilizes deep learning. However, it is preferable that the detection algorithm of the second detection section 14 be able to detect with higher accuracy and have a larger amount of calculation than the detection algorithm of the first detection section 12.

[0038] The output unit 15 outputs detection results for the objects detected by the second detection unit 14. For example, the output unit 15 outputs result information indicating that an object has been detected for candidate regions whose reliability based on the detection results of the second detection unit 14 is above a threshold. Candidate regions with a reliability below the threshold may not be included in the result information. The detection result information is not particularly limited; for example, in the case of face detection, information such as face region, reliability, face orientation, age, gender, ethnicity, and expression may be included.

[0039] <Processing Content>

[0040] Figure 3 This is a flowchart illustrating the overall process of object detection processing based on the object detection device 10. Hereinafter, according to... Figure 3 The flowchart describes the details of the object detection device 10.

[0041] S31: Image Input Processing

[0042] In step S31, the object detection device 10 acquires an image (input image). The input image can be acquired from the camera 20 via the image input unit 11, from another computer via the communication device 104, or from the storage unit 16.

[0043] S32: First Detection Processing

[0044] In step S32, the first detection unit 12 detects candidate regions (regions where objects presumed to be the target of detection exist) from the input image (first detection process). In this embodiment, the first detection unit 12 is configured to use Haar-like features as image features and AdaBoost as the learning algorithm. The detection result of the first detection process may include, in addition to the aforementioned candidate regions, the certainty that an object being detected exists within that candidate region (first detection reliability, detection score).

[0045] S33: Decision Processing

[0046] In step S33, the determination unit 13 determines the region other than the candidate regions detected in step S32 that have a similarity to the false detection region of a given value or higher as the object region. The false detection region is an object region in one or more previous frames where no object was detected in the second detection process described later. The determination unit 13 outputs the region after removing regions similar to the false detection region from the candidate regions detected in step S32 as the object region.

[0047] use Figure 4 The determination process performed in step S33 is explained in detail. Figure 4is a flowchart of the determination processing involved in the present embodiment. First, the determination section 13 acquires the detection information (positions and sizes of the false detection regions) from the storage section 16 (S41). The determination section 13 can acquire only the false detection information for the immediately preceding frame, or can acquire the false detection information for a given number of most recent frames. Then, the determination section 13 calculates the similarity to the false detection region for each of one or more candidate regions (S42). In the present embodiment, as an index of the similarity of regions to each other, the IoU (Intersection over Union) is used. The IoU is a value obtained by dividing the area of the intersection of two regions by the area of the union of the two regions. The IoU takes a value between 0 and 1, and is 1 if the two regions completely overlap, and is 0 if they do not overlap at all. In the calculation of the IoU, the positions and sizes of the candidate regions, and the positions and sizes of the false detection regions can be used. Then, the determination section 13 determines whether the IoU is equal to or greater than a given threshold value T1 (S43), and in the case where the IoU is equal to or greater than the threshold value T1, outputs the regions excluding the corresponding candidate region as the object region (S44).

[0048] S34 to S36: Second Detection Processing

[0049] In step S34, the second detection section 14 determines whether the one or more object regions output in step S33 include the object to be detected (second detection processing). In the present embodiment, the second detection section 14 performs object detection using an identifier that has been learned using a multi-layer neural network called a convolutional neural network (CNN).

[0050] In step S35, the second detection section 14 determines whether there is an object region that was determined in the processing of step S34 to not include the object to be detected.

[0051] In step S36, the second detection section 14 records information related to the object region determined to not include the object to be detected in the storage section 16 as detection information. In the present embodiment, as the detection information, the position and size of the object region determined to not include the object to be detected are recorded in the storage section 16.

[0052] S37: Detection Result Output Processing

[0053] In step S37, the output section 15 outputs a detection result for the region in which the object was detected in step S34. The output section 15 outputs result information indicating that the object to be detected was detected, for the detection target region for which the reliability of the detection result based on the detection by the second detection section 14 (second detection reliability) is equal to or greater than a threshold value. The detection target region for which the reliability is less than the threshold value can not be included in the result information.

[0054] <Advantages of the Present Embodiment>

[0055] In object detection, false detection of the same object in an input image can occur multiple times, but according to the present embodiment, false detection of the same position and size of an object multiple times can be effectively reduced. Thus, the number of candidate regions (object regions) submitted to the second detection section is reduced, and therefore, by performing two-stage detection processing, the processing time can be reduced while maintaining the detection performance.

[0056] (Embodiment 2)

[0057] In Embodiment 1 described above, an example in which the degree of similarity is decided based on the position and size of the candidate region and the false detection region in step S33 is described. In the present embodiment, an example in which the degree of similarity is decided by performing pattern matching of an image corresponding to the candidate region and an image corresponding to the false detection region in step S33 is described. The same processing as in Embodiment 1 described above is omitted from the description, and the different processing, i.e., the determination processing (S33) is described.

[0058] <Determination Processing (S33)>

[0059] Figure 5 Fig. 13 is a flowchart of the determination processing performed in step S33 in the present embodiment. First, the determination section 13 acquires the detection information from the storage section 16 (S51). In the present embodiment, an image corresponding to the false detection region is included in the detection information. Then, the determination section 13 performs pattern matching processing using the image corresponding to the false detection region for each of the images corresponding to one or more candidate regions (S52). Then, the determination section 13 determines whether the degrees of similarity of the images obtained by the pattern matching to each other are equal to or greater than a given threshold value T2 (S53), and in the case where the degrees of similarity are equal to or greater than the threshold value T2, outputs the region excluding the corresponding candidate region as an object region (S54).

[0060] <Advantages of the Present Embodiment>

[0061] According to the present embodiment, object detection can be performed with high accuracy even in the case where the position and size of the false detection region and the candidate region are identical or similar, but the images corresponding to the two regions are completely different. For example, in the case where the object 103 is detected in the position of the object 103 shown in Fig. 12, even in the case where the detected object overlaps, the degree of similarity can be calculated based on the images, and therefore, the region corresponding to the position can be output as an object region. Figure 1

[0062] (Variation)

[0063] ​In Embodiment 1 and Embodiment 2 described above, an example in which the determination section 13 determines the candidate region other than the region similar to the false detection region as the object region is described, but is not limited thereto. For example, in a case where the first detection section 12 outputs the first detection reliability described above, the determination section 13 determines the candidate region in which the first detection reliability is equal to or higher than a given threshold value T3 as the object region. At this time, the determination section 13 can also determine, with respect to the candidate region in which the degree of similarity to the false detection region is equal to or higher than a given threshold value T4, the candidate region in which a value obtained by subtracting a given value from the first detection reliability is equal to or higher than the given threshold value T3 described above as the object region.

[0064] In addition, a method of determining the given value to be subtracted from the first detection reliability is not particularly limited. The given value to be subtracted from the reliability can also be a fixed value. Furthermore, the given value to be subtracted from the reliability can also be determined in accordance with the number of consecutive frames in which the object is not detected by the second detection section 14. For example, the given value to be subtracted from the first detection reliability can be increased as the number of consecutive frames increases, or the given value to be subtracted from the first detection reliability can be subtracted for the first time in a case where the number of consecutive frames is equal to or higher than a certain number. Furthermore, in a case where the second detection reliability is output by the second detection section 14, the given value to be subtracted from the reliability can also be determined on the basis of the second detection reliability. For example, the determination section 13 determines the candidate region in which the first detection reliability is equal to or higher than a given threshold value T3 as the object region. At this time, the determination section 13 can also determine, with respect to the candidate region in which the degree of similarity to the false detection region is equal to or higher than a given threshold value T4, the candidate region in which a value obtained by subtracting a value on the basis of the second detection reliability from the first detection reliability is equal to or higher than the given threshold value T3 described above as the object region. For example, the higher the second detection reliability, the larger the given value to be subtracted from the reliability can be.

[0065] In Embodiment 1 described above, an example in which the IoU is used as an index of the degree of similarity between regions is described, but is not limited thereto. For example, a ratio or a difference in size between regions, or a difference in position (for example, a coordinate value of the center) between regions, or a combination thereof can be used as an index of the degree of similarity.

[0066] In Embodiment 2 described above, an example in which pattern matching is performed when the degree of similarity between images is calculated is described, but is not limited thereto. For example, a difference in color information, a difference in luminance information in the images can be used as an index of the degree of similarity.

[0067] Explanation of Reference Signs

[0068] 10: object detection device

[0069] 11: image input section

[0070] 12: first detection section

[0071] 13: determination section

[0072] 14: second detection section

[0073] 15: output section

[0074] 16: storage section

[0075] 1, 20: camera

[0076] 101, 102, 103: object

[0077] 111, 112, 113: candidate region

[0078] 121, 122, 123: object region

Claims

1. An object detection device for detecting a given object from an image, characterized in that it comprises: The first detection unit detects candidate regions where the object exists from the image; The determination unit determines the object region from one or more candidate regions detected by the first detection unit; The second detection unit takes the object region as the object and detects the object using a different detection algorithm than the first detection unit; as well as The storage unit stores detection information representing the detection results of the second detection unit for the object region. The determination unit determines the object region from one or more candidate regions based on the detection information from one or more previous frames. The detection information includes information related to object regions where the object was not detected by the second detection unit. The first detection unit also outputs a first detection reliability representing the certainty that the object is contained in the candidate region. The determination unit determines the object region based on a value obtained by subtracting a given value from the first detection reliability for candidate regions that have a similarity of more than a given value to object regions where the object was not detected in the previous frame. For other candidate regions, the determination of the object region is based on the first detection reliability.

2. The object detection device according to claim 1, characterized in that, The given value is a value corresponding to the number of consecutive frames in which the object was not detected by the second detection unit.

3. The object detection device according to claim 1, characterized in that, The given value is a fixed value.

4. The object detection device according to any one of claims 1 to 3, characterized in that, The detection information includes the location and / or size of the object region. The determination unit calculates the similarity based on the position and / or size of the candidate region and the position and / or size of the object region.

5. The object detection device according to any one of claims 1 to 3, characterized in that, The detection information includes the image corresponding to the object region. The determination unit calculates the similarity based on the image contained in the detection information and the image corresponding to the candidate region.

6. An object detection apparatus for detecting a given object from an image, the object detection apparatus being characterized by having: The first detection unit detects candidate regions where the object exists from the image; The determination unit determines the object region from one or more candidate regions detected by the first detection unit; The second detection unit takes the object region as the object and detects the object using a different detection algorithm than the first detection unit; as well as The storage unit stores detection information representing the detection results of the second detection unit for the object region. The first detection unit also outputs a first detection reliability representing the certainty that the object is contained in the candidate region. The detection information includes a second detection reliability, determined by the second detection unit, which represents the certainty that the object is contained in the object region. The determination unit determines the object region based on a value obtained by subtracting the value corresponding to the second detection reliability from the first detection reliability for candidate regions whose similarity to the object region shown in the detection information is greater than or equal to a given value. For other candidate regions, the determination of the object region is based on the first detection reliability.

7. The object detection device according to claim 6, characterized in that, The detection information includes the location and / or size of the object region. The determination unit calculates the similarity based on the position and / or size of the candidate region and the position and / or size of the object region.

8. The object detection device according to claim 6, characterized in that, The detection information includes the image corresponding to the object region. The determination unit calculates the similarity based on the image contained in the detection information and the image corresponding to the candidate region.

9. An object detection method for detecting a given object from an image, characterized in that it has: The first detection step is to detect candidate regions where the object exists from the image; The determination step involves deciding on the object region from one or more candidate regions detected in the first detection step. The second detection step involves taking the object region as the object and detecting the object using a different detection algorithm than the first detection step. as well as The storage step stores detection information representing the detection results of the second detection step for the object region. In the determination step, the object region is determined from one or more candidate regions based on the detection information from one or more previous frames. The detection information includes information related to object regions where the object was not detected through the second detection step. In the first detection step, a first detection reliability score, representing the certainty that the object is contained in the candidate region, is also output. In the determination step, for candidate regions whose similarity to object regions where the object was not detected in the previous frame is greater than or equal to a given value, the object region is determined based on the value obtained by subtracting the given value from the first detection reliability. For other candidate regions, the object region is determined based on the first detection reliability.

10. A recording medium containing a program for causing a computer to perform the steps of the object detection method of claim 9.

Images