JP2022146384A - Object detection device - Google Patents

Abstract

To provide an object detection device with which, even when a plurality of objects overlap in an image, it is possible to detect each of these objects.SOLUTION: The object detection device comprises: an object detection unit 31 for detecting a plurality of object areas from an image, in which objects to be detected are displayed, as well as calculating, for each object area, hide reliability that represents the likelihood of an object being hidden at least partly by other objects; an overlap assessment unit 33 which, when, regarding a set of object areas overlapping each other at least in part, the overlap degree of object areas is greater than or equal to a prescribed overlap degree threshold, assesses that one object to be detected is represented in this set of object areas, and which, when the overlap degree is less than the prescribed overlap degree threshold, assesses that a separate object to be detected for each object area is represented; and a threshold setting unit 32 for raising the prescribed overlap degree threshold progressively as the difference of hide reliability between the object areas included in a set of object areas increases.SELECTED DRAWING: Figure 3

Description

The present invention relates to an object detection device that detects an object represented in an image.

Techniques for detecting a specific object represented in an image obtained by a camera have been researched (see, for example, Patent Document 1).

The image processing device disclosed in Patent Literature 1 detects an object from an image and also detects a specific object from the image. This video processing device determines whether the detected object and the detected specific object correspond based on the ratio of overlap between the detected object area and the detected specific object area. At this time, the image processing device determines that the object and the specific object correspond when the ratio of overlap between the detected object area and the detected specific object area exceeds a predetermined threshold.

JP 2015-46194 A

In the above technique, objects detected by different detection means and specific objects are associated with each other. However, multiple objects to be detected may be represented on the image, one of which may be partially obscured by the other. In such cases, it may be difficult to detect each object with the techniques described above.

SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide an object detection apparatus capable of detecting each object even when a plurality of objects overlap each other on an image.

According to one embodiment, an object detection device is provided. This object detection apparatus inputs an image representing a predetermined area generated by an imaging unit to a classifier, thereby detecting a plurality of object areas representing objects to be detected from the image, and detecting a plurality of object areas. an object detection unit that calculates, for each of the object regions, a hidden reliability that indicates the likelihood that an object to be detected is at least partially hidden by another object; For a set of partially overlapping object regions, if the degree of overlap between object regions included in the set of object regions is equal to or greater than a predetermined degree of overlap threshold, one object to be detected is present in the set of object regions. Overlapping determination determining that each object region included in the set of object regions represents a separate object to be detected when the degree of overlap is less than a predetermined degree of overlap threshold. and a threshold value setting unit that increases a predetermined overlapping degree threshold as the difference in hidden reliability between object regions included in the set of object regions increases.

INDUSTRIAL APPLICABILITY The object detection apparatus according to the present invention has the effect of being able to detect each object even when a plurality of objects overlap on an image.

1 is a schematic configuration diagram of a vehicle control system in which an object detection device is mounted; FIG. 1 is a hardware configuration diagram of an electronic control device that is an embodiment of an object detection device; FIG. FIG. 4 is a functional block diagram of a processor of the electronic control unit regarding vehicle control processing including object detection processing; FIG. 2 is a diagram showing an example of the configuration of a DNN used as a discriminator; FIG. FIG. 10 is a diagram showing an example of an image showing a plurality of detection target objects; 4 is an operation flowchart of vehicle control processing including object detection processing;

The object detection device will be described below with reference to the drawings. This object detection device inputs an image to a classifier trained in advance so as to detect one or more objects to be detected (hereinafter sometimes referred to as detection target objects), and detects on the image. A plurality of regions including the target object (hereinafter sometimes referred to as object regions) are detected. In this object detection apparatus, for a set of object areas that at least partially overlap each other, if the degree of overlap between the object areas is equal to or greater than a predetermined overlap degree threshold value, one detection target object appears in the set of object areas. It is determined that On the other hand, if the degree of overlap is less than the degree of overlap threshold, the object detection device determines that each of the object regions included in the set of object regions represents a separate detection target object. Here, the discriminator further calculates, for each object region, a hidden reliability that indicates the probability that the detection target object represented in the object region is at least partially hidden by another object. Then, the object detection apparatus increases the predetermined overlap degree threshold as the difference in the hidden reliability of the object areas included in the set of object areas that at least partially overlap each other increases. Thus, this object detection device accurately determines whether the discriminator has detected a plurality of object regions for one detection target object or whether the object regions have been detected separately for a plurality of detection target objects. make it possible. As a result, this object detection device makes it possible to detect each of multiple objects even when the objects overlap on the image.

An example in which the object detection device is applied to a vehicle control system will be described below. In this example, the object detection device detects a detection target object existing around the vehicle by performing object detection processing on a series of time-series images obtained by a camera mounted on the vehicle. Objects to be detected include objects that affect the traveling of the vehicle 10, such as other vehicles traveling around the vehicle 10, people, road signs, traffic lights, road markings such as lane markings, and others on the road. objects such as

FIG. 1 is a schematic configuration diagram of a vehicle control system in which an object detection device is mounted. Also, FIG. 2 is a hardware configuration diagram of an electronic control unit, which is one embodiment of the object detection apparatus. In this embodiment, a vehicle control system 1 mounted on a vehicle 10 and controlling the vehicle 10 includes a camera 2 for capturing an image of the surroundings of the vehicle 10 and an electronic control unit (ECU), which is an example of an object detection device. 3. The camera 2 and the ECU 3 are communicably connected via an in-vehicle network conforming to a standard such as a controller area network. The vehicle control system 1 may further include a storage device that stores map information representing the position and type of features, lane markings, and the like, which are used for automatic driving control of the vehicle 10 . Further, the vehicle control system 1 includes a distance measuring sensor such as LiDAR or radar, a receiver such as a GPS receiver for measuring the self-position of the vehicle 10 in compliance with the satellite positioning system, and a receiver for wirelessly communicating with other devices. It may have a wireless terminal. Furthermore, the vehicle control system 1 may have a navigation device for searching for the planned travel route of the vehicle 10 .

The camera 2 is an example of an imaging unit, and includes a two-dimensional detector composed of an array of photoelectric conversion elements sensitive to visible light, such as a CCD or C-MOS, and an object to be photographed on the two-dimensional detector. It has an imaging optical system that forms an image of the area. The camera 2 is installed, for example, in the vehicle interior of the vehicle 10 so as to face the front of the vehicle 10 . Then, the camera 2 photographs the area in front of the vehicle 10 at predetermined photographing intervals (for example, 1/30 second to 1/10 second) and generates an image showing the area in front of the vehicle. The image obtained by camera 2 is preferably a color image. Note that the vehicle 10 may be provided with a plurality of cameras having different photographing directions or focal lengths.

Each time the camera 2 generates an image, it outputs the generated image to the ECU 3 via the in-vehicle network.

The ECU 3 controls the vehicle 10 . In this embodiment, the ECU 3 controls the vehicle 10 to automatically drive the vehicle 10 based on an object detected from a series of time-series images obtained by the camera 2 . Therefore, the ECU 3 has a communication interface 21 , a memory 22 and a processor 23 .

The communication interface 21 is an example of a communication section, and has an interface circuit for connecting the ECU 3 to the in-vehicle network. That is, the communication interface 21 is connected to the camera 2 via the in-vehicle network. Then, the communication interface 21 passes the received image to the processor 23 every time it receives an image from the camera 2 .

The memory 22 is an example of a storage unit, and has, for example, a volatile semiconductor memory and a nonvolatile semiconductor memory. The memory 22 stores various data and parameters used in vehicle control processing including object detection processing executed by the processor 23 of the ECU 3 . As such data or parameters, the memory 22 stores, for example, the image received from the camera 2, various parameters for specifying classifiers used in object detection processing, confidence thresholds for each type of object, and the like. Remember. Furthermore, the memory 22 stores various data generated during the vehicle control process for a certain period of time. Furthermore, the memory 22 may store information used for driving control of the vehicle 10, such as map information.

The processor 23 is an example of a control unit, and has one or more CPUs (Central Processing Units) and their peripheral circuits. The processor 23 may further comprise other arithmetic circuits such as a logic operation unit, a numerical operation unit or a graphics processing unit (GPU). Each time an image is received from the camera 2 while the vehicle 10 is running, the processor 23 executes vehicle control processing including object detection processing on the received image. The processor 23 then controls the vehicle 10 to automatically drive the vehicle 10 based on the detected objects around the vehicle 10 .

FIG. 3 is a functional block diagram of the processor 23 of the ECU 3 regarding vehicle control processing including object detection processing. The processor 23 has an object detection section 31 , a threshold setting section 32 , an overlap determination section 33 , a driving planning section 34 and a vehicle control section 35 . These units of the processor 23 are, for example, functional modules implemented by computer programs running on the processor 23 . Alternatively, each of these units of processor 23 may be a dedicated arithmetic circuit provided in processor 23 . Among these units of the processor 23, the object detection unit 31, the threshold value setting unit 32, and the duplication determination unit 33 execute object detection processing. In addition, when a plurality of cameras are provided in the vehicle 10, the processor 23 may perform object detection processing based on the image obtained by each camera.

Each time an image is received from the camera 2, the object detection unit 31 inputs the received latest image to the classifier for object detection, thereby detecting an object area including the detection target object represented in the image. Also, the type of the object to be detected is specified. Further, the object detection unit 31 calculates the hidden reliability for each object area.

The object detection unit 31 detects an object region including a detection target object represented in an image as a classifier, identifies the type of the detection target object, and uses a pre-trained DNN to calculate a hidden reliability. take advantage of The DNN used by the object detection unit 31 can be, for example, a DNN having a convolutional neural network (hereinafter simply referred to as CNN) type architecture.

FIG. 4 is a diagram showing an example of the configuration of a DNN used as a discriminator. The DNN 400 has a main body 401 provided on the input side to which an image is input, and a position detection section 402 , a type estimation section 403 and a hidden reliability calculation section 404 provided on the output side of the main body 401 . Based on the output from the main body 401, the position detection unit 402 outputs the circumscribed rectangle of the detection target object represented on the image as the object region. Based on the output from the main body 401 , the type estimating section 403 calculates a certainty factor for each type of detection target object represented in the object region detected by the position detecting section 402 . Then, the hidden reliability calculation unit 404 calculates the hidden reliability of each object region based on the output from the main body 401 . Two or more of the position detection unit 402, the type estimation unit 403, and the hidden reliability calculation unit 404 may be integrally formed.

The main trunk 401 can be, for example, a CNN with multiple layers connected in series from the input side to the output side. The multiple layers include two or more convolutional layers. Furthermore, the plurality of layers included in main trunk 401 may include pooling layers provided for each of one or more convolution layers.

When an image is input, the main body 401 outputs a feature map calculated from the image by executing operations in each layer on the image. Note that the main body 401 may output a plurality of feature maps with different resolutions. For example, the main body 401 may output a feature map having the same resolution as the input image and one or more feature maps having a lower resolution than the input image.

The feature map output from the main body 401 is input to the position detection unit 402, the type estimation unit 403, and the hidden reliability calculation unit 404, respectively. The position detection unit 402, the type estimation unit 403, and the hidden reliability calculation unit 404 can each be, for example, a CNN having multiple layers connected in series from the input side to the output side. For each of the position detection unit 402, the type estimation unit 403, and the hidden reliability calculation unit 404, the multiple layers of the CNN include two or more convolution layers. Further, for each of the position detection unit 402, the type estimation unit 403, and the hidden reliability calculation unit 404, the multiple layers of the CNN may include a pooling layer provided for each one or multiple convolution layers. Note that the convolution layer and pooling layer of the CNN may be shared by the position detection unit 402 , the type estimation unit 403 and the hidden reliability calculation unit 404 . Furthermore, for each of the position detection unit 402, the type estimation unit 403, and the hidden reliability calculation unit 404, the plurality of layers may include one or more fully connected layers. In this case, the fully connected layer is preferably provided on the output side of each convolutional layer. Alternatively, the output from each convolutional layer may be directly input to the fully connected layer. The output layer of the type estimating unit 403 may be a softmax layer that calculates the confidence of each type of the detection target object according to the softmax function, or a softmax layer that calculates the confidence of each type of the detection target object according to the sigmoid function. It may be a sigmoid layer to be calculated. Furthermore, the output layer of the hidden reliability calculation unit 404 can be a sigmoid layer that calculates the hidden reliability of each object region according to a sigmoid function.

The position detecting unit 402 and the type estimating unit 403 are trained to output respective degrees of certainty of the type of the detection target object, for example, for regions of various positions on the image, various sizes, and various aspect ratios. be. Therefore, when an image is input, the discriminator 400 outputs respective confidence factors of the detection target object type for each region of various positions, various sizes, and various aspect ratios on the image. Then, the position detecting unit 402 and the type estimating unit 403 detect an area where the certainty of any type of detection target object is equal to or higher than a predetermined certainty threshold as an object area representing that type of detection target object. do.

On the other hand, the hidden reliability calculation unit 404 calculates, for example, a hidden value representing the probability that the detection target object is hidden by another object for each object region having various positions, sizes, and aspect ratios on the image. It is learned to calculate reliability. Therefore, when a plurality of detection target objects are represented on the image, the discriminator detects an object region for each detection target object and calculates the hidden reliability for each object region.

An image (teacher image) included in the training data used for learning of the discriminator 400 represents, for example, the type of detection target object (eg, ordinary passenger car, bus, truck, motorcycle, etc.) and the detection target object. The circumscribing rectangle of the object to be detected, which is the object region that is detected, is tagged. Furthermore, each object area on the teacher image is tagged with the degree to which the detection target object included in the object area is hidden by other objects.

The discriminator 400 is trained using a large number of teacher images as described above, for example, according to a learning technique such as error backpropagation. By using the classifier 400 learned in this way, the processor 23 can accurately detect the object to be detected from the image. Further, the discriminator 400 can accurately calculate the hidden reliability for each object region.

The object detection unit 31 registers the position and range of each object area on the image, the type of object included in the object area, and the degree of hidden reliability in the detected object list. The object detection unit 31 then stores the detected object list in the memory 22 .

The threshold setting unit 32 refers to the detected object list to identify a set of object regions that at least partially overlap each other. Then, the threshold setting unit 32 sets an overlap degree threshold for determining whether or not the set represents the same detection target object for each identified set of object regions. In the present embodiment, the threshold setting unit 32 increases the degree of overlap threshold as the difference in hidden reliability between two object regions included in the identified pair increases, and different objects appear in the two object regions. make it easier to determine that

For example, the threshold setting unit 32 sets the relational expression between the hidden reliability difference and the overlapping degree threshold so that the greater the hidden reliability difference, the higher the overlapping degree threshold. By inputting the hidden reliability difference of the regions, the degree of overlap threshold for the pair is set. Alternatively, the threshold setting unit 32 may set the overlap degree threshold to the first value when the difference in the hidden reliability between two object regions included in the set of object regions of interest is greater than a predetermined difference threshold. good. On the other hand, if the difference in hidden reliability is equal to or less than the predetermined difference threshold, the threshold setting unit 32 may set the overlap degree threshold to a second value lower than the first value.

The threshold setting unit 32 notifies the overlap determination unit 33 of the overlap degree threshold set for each of the sets of object regions that at least partially overlap each other.

The overlap determination unit 33 performs non-maximum suppression (NMS) processing on each of pairs of object regions that at least partially overlap with each other, thereby determining whether or not the same detection target object is represented in the pairs. judge. Then, if the duplication determination unit 33 determines that the same detection target object is represented, it selects one object region (for example, the larger object region) from the set, and selects the non-selected object region. Delete the object area and the detection target object from the detection object list.

Specifically, the overlap determination unit 33 calculates the degree of overlap between object regions included in a set of object regions of interest, and compares the calculated degree of overlap with the degree of overlap threshold applied to the set. For example, the overlap determination unit 33 determines, as the degree of overlap of a set of object regions of interest, the ratio of the area of overlapping regions to the area of the entire set of object regions included in the set (Intersection over Union (IoU)). Calculate Alternatively, the overlap determination unit 33 may calculate the ratio of the area of the overlapping area to the area of the larger object area as the degree of overlap of the set of object areas of interest. Then, the overlap determination unit 33 determines that the same detection target object is represented for pairs whose degree of overlap is equal to or greater than the degree of overlap threshold among pairs of object regions that at least partially overlap each other. On the other hand, the overlap determination unit 33 determines that different detection target objects are represented for each of the object regions whose degree of overlap is less than the degree of overlap threshold among the pairs of object regions that at least partially overlap each other. .

FIG. 5 is a diagram showing an example of an image showing a plurality of detection target objects. An image 500 shows a plurality of vehicles, which are objects to be detected. It is assumed that the vehicles 501 to 503 are detected by the object detection unit 31 among the plurality of vehicles. Assume that the object detection unit 31 sets an object area 511 including the vehicle 501 , an object area 512 including the vehicle 502 , and two object areas 513 and 514 including the vehicle 503 . Also, the object region 511 and the object region 512 partially overlap. Similarly, object region 513 and object region 514 partially overlap.

Since the vehicle 501 is not hidden by other objects, the hidden reliability calculated for the object region 511 including the vehicle 501 is a relatively small value. In contrast, vehicle 502 is partially hidden by vehicle 501 . Therefore, the hidden reliability calculated for the object region 512 including the vehicle 502 is a relatively large value. Therefore, the difference between the hidden reliability calculated for the object region 511 and the hidden reliability calculated for the object region 512 is also a relatively large value. also higher. As a result, it becomes easier to accurately determine that the vehicle 501 represented in the object region 511 and the vehicle 502 represented in the object region 512 are different objects.

On the other hand, since the object regions 513 and 514 are set for the same vehicle 503 , the hidden reliability of the object region 513 is approximately the same as the hidden reliability of the object region 514 . Therefore, the overlapping degree threshold applied to the set of object regions 513 and 514 is low. As a result, it becomes easier to accurately determine that the object regions 513 and 514 include the same vehicle 503 .

Note that when the types of objects represented by the object regions included in the set of object regions that at least partially overlap each other are different, the overlap determination unit 33 determines that separate detection target objects are present in the respective object regions. It may be determined that the

The operation planning unit 34 refers to the detected object list and determines a predetermined section from the current position of the vehicle 10 to a predetermined distance (for example, 500 m to 1 km) so that the vehicle 10 does not collide with objects around the vehicle 10. One or more planned travel routes for the vehicle 10 in are generated. The planned travel route is represented, for example, as a set of target positions of the vehicle 10 at each time when the vehicle 10 travels a predetermined section.

In order to generate a planned travel route, the operation planning unit 34 tracks a detection target object (hereinafter simply referred to as an object) registered in the detection object list, predict the trajectory of an object

For example, the operation planning unit 34 applies tracking processing based on optical flow, such as the Lucas-Kanade method, to object regions of interest in the latest image obtained by the camera 2 and object regions in past images. to track the object represented in the object region. Therefore, the operation planning unit 34 extracts a plurality of feature points from the object region of interest by, for example, applying a feature point extraction filter such as SIFT or Harris operator to the object region of interest. Then, the operation planning unit 34 may calculate the optical flow by specifying the corresponding point in the object region in the past image for each of the plurality of feature points according to the applied tracking method. Alternatively, the operation planning unit 34 may apply another tracking method applied to tracking a moving object detected from an image to the object region of interest in the latest image and the object region in the previous image. , may track the object represented in that object region.

For each object being tracked, the operation planning unit 34 performs viewpoint conversion processing using information such as the mounting position of the camera 2 on the vehicle 10, thereby converting the in-image coordinates of the object to the coordinates on the bird's-eye view image. (bird's-eye coordinates). At that time, the operation planning unit 34 determines the position and orientation of the vehicle 10 at the time of acquisition of each image, the estimated distance to the detected object, and the direction from the vehicle 10 toward the object, and determines the , the position of the detected object can be estimated. Note that the driving planning unit 34 obtains the position and attitude of the vehicle 10 at the time each image is acquired, for example, from a GPS receiver (not shown) mounted on the vehicle 10, and presents the current position information representing the current position of the vehicle 10. can be estimated based on Alternatively, every time an image is obtained by the camera 2, the driving planning unit 34 detects the left and right lane markings of the vehicle 10 from the image, and compares the detected lane markings with the map information stored in the memory 22. The position and orientation of the vehicle 10 may be estimated by matching . Also, the driving planning unit 34 can specify the direction from the vehicle 10 to the object based on the position of the object area including the detected object on the image and the optical axis direction of the camera 2 . Furthermore, it is assumed that the position of the lower end of the object region represents the position of the object represented in the object region that touches the road surface. Therefore, the driving planning unit 34 can estimate the distance to the object represented in the object area based on the azimuth from the camera 2 corresponding to the lower end of the object area and the installation height of the camera 2 . Then, the operation planning unit 34 executes prediction processing using a Kalman filter, a particle filter, or the like for a series of bird's-eye coordinates in the most recent predetermined period, thereby estimating the predicted trajectory of the object up to a predetermined time ahead. can be done.

Based on the predicted trajectory of each object being tracked, the operation planning unit 34 predicts the distance between each object being tracked and the vehicle 10 up to a predetermined time ahead for any object is a predetermined distance or more, and , the planned travel route of the vehicle 10 is set along the planned travel route to the destination. At that time, the operation planning unit 34 calculates, as an evaluation function, the reciprocal of the sum of the distances to the object being tracked that is closest to the position on the planned travel route at each time up to a predetermined time ahead. Then, the operation planning unit 34 may set a planned travel route according to a predetermined optimization method such as dynamic programming or steepest descent so that the evaluation function is minimized.
Note that the operation planning unit 34 may generate a plurality of planned travel routes. In this case, the operation planning unit 34 may select the route that minimizes the sum of the absolute values of the acceleration of the vehicle 10 from among the plurality of planned travel routes.

The operation planning unit 34 notifies the vehicle control unit 35 of the generated planned travel route.

The vehicle control unit 35 controls each part of the vehicle 10 so that the vehicle 10 travels along the notified planned travel route. For example, the vehicle control unit 35 obtains the target acceleration of the vehicle 10 according to the notified planned travel route and the current vehicle speed of the vehicle 10 measured by a vehicle speed sensor (not shown), and calculates the target acceleration. Set the accelerator opening or brake amount to . The vehicle control unit 35 obtains the fuel injection amount according to the set accelerator opening, and outputs a control signal corresponding to the fuel injection amount to the fuel injection device of the engine of the vehicle 10 . Alternatively, the vehicle control unit 35 outputs a control signal corresponding to the set brake amount to the brake of the vehicle 10 .

Further, when the vehicle 10 changes its course so that the vehicle 10 travels along the planned travel route, the vehicle control unit 35 obtains the steering angle of the vehicle 10 according to the planned travel route. The vehicle control unit 35 then outputs a control signal corresponding to the steering angle to an actuator (not shown) that controls the steered wheels of the vehicle 10 .

FIG. 6 is an operation flowchart of vehicle control processing including object detection processing executed by the processor 23 . Each time the processor 23 receives an image from the camera 2, it executes vehicle control processing according to the operation flowchart shown in FIG. In the operation flowchart shown below, the processing of steps S101 to S106 corresponds to the object detection processing.

The object detection unit 31 of the processor 23 inputs the image obtained from the camera 2 to the classifier, and detects one or more detection target objects represented in the image. That is, the object detection unit 31 detects one or more object regions including the detection target object on the image (step S101). Furthermore, the object detection unit 31 identifies the type of each detected detection target object. Then, the object detection unit 31 registers the detected object to be detected in the detected object list. Furthermore, the object detection unit 31 calculates the hidden reliability of each detected object area (step S102).

The threshold setting unit 32 of the processor 23 sets the overlap degree threshold so that the overlap degree threshold increases as the difference in the hidden reliability increases for a set of object regions that at least partially overlap each other (step S103). Then, the overlap determination unit 33 of the processor 23 determines whether or not the degree of overlap of a set of object regions that at least partially overlap each other is equal to or greater than the degree of overlap threshold (step S104).

If the overlap degree is equal to or greater than the overlap degree threshold value (step S104-Yes), the overlap determination unit 33 determines that the same detection target object is represented in the set of object regions that at least partially overlap each other. The duplication determination unit 33 then deletes one of the object regions from the detected object list (step S105).

On the other hand, if the overlap degree is less than the overlap degree threshold value (step S104-No), the overlap determination unit 33 determines that each object region represents a separate detection target object (step S106).

After step S105 or step S106, the operation planning unit 34 of the processor 23 refers to the detected object list and tracks each detection target object registered in the detected object list to estimate the predicted trajectory of the object. . Then, the operation planning unit 34 generates a planned travel route for the vehicle 10 so that the estimated predicted trajectory is equal to or greater than a predetermined distance (step S107). The vehicle control unit 35 of the processor 23 controls the vehicle 10 so that the vehicle 10 travels along the planned travel route (step S108). The processor 23 then terminates the vehicle control process.

As described above, this object detection apparatus selects one detection target object according to the degree of overlap for a set of object regions that at least partially overlap with each other, detected by inputting an image to the classifier. is represented or a separate detectable object is represented. At this time, the object detection apparatus increases a predetermined degree of overlap threshold to be compared with the degree of overlap as the difference in the degree of hidden reliability between the object regions included in the set of object regions increases. Thus, this object detection device accurately determines whether the discriminator has detected a plurality of object regions for one detection target object or whether the object regions have been detected separately for a plurality of detection target objects. be able to. As a result, this object detection apparatus can detect each of multiple objects even when the objects overlap on the image.

The object detection device according to the above embodiment or modification may be mounted in a device other than an in-vehicle device. For example, the object detection device according to the above embodiment or modification detects a predetermined object from an image generated by a surveillance camera installed to photograph a predetermined outdoor or indoor area at predetermined intervals. may be configured. Then, when a predetermined object is detected over a certain period of time, the object detection device may cause a display connected to the object detection device to display a message indicating that the predetermined object has been detected.

In addition, the computer program that implements the function of each unit of the processor 23 of the object detection device according to the above embodiment or modification can be stored in a computer-readable portable recording medium such as a semiconductor memory, a magnetic recording medium, or an optical recording medium. May be provided in recorded form.

As described above, those skilled in the art can make various modifications within the scope of the present invention according to the embodiment.

1 vehicle control system 2 camera 3 electronic control device (object detection device)
21 communication interface 22 memory 23 processor 31 object detection unit 32 threshold setting unit 33 duplication determination unit 34 operation planning unit 35 vehicle control unit

Claims (1)

By inputting an image representing a predetermined area generated by the imaging unit to the classifier, a plurality of object areas representing objects to be detected are detected from the image, and each of the plurality of object areas is detected. an object detection unit that calculates a hidden reliability representing the likelihood that the object to be detected is at least partially hidden by another object;
For a set of object regions that at least partially overlap each other among the plurality of detected object regions, if the degree of overlap between object regions included in the set of object regions is equal to or greater than a predetermined degree of overlap threshold, If it is determined that one object to be detected is represented in a set of object regions, and the degree of overlap is less than the predetermined degree of overlap threshold, each object region included in the set of object regions is separately detected. A duplication determination unit that determines that the object to be detected is represented in
a threshold setting unit configured to increase the predetermined overlapping degree threshold as the difference in the hidden reliability between the object regions included in the set of object regions increases;
An object detection device having

Images