KR102243493B1 - System for enforcing bad and overloading vehicle to prevent falling object in cargo truck - Google Patents

Abstract

The present invention relates to an illegal loading enforcement system, which comprises: a photographing device photographing the rear surface of a cargo truck passing on a road; an image collection device collecting images captured by the photographing device; an image processing device including a load classification engine that classifies the cargo truck by performing a learning process using deep learning on a rear image of the cargo truck collected by the image collection device, classifying a target cargo truck through the load classification engine when the image of the target cargo truck is input from the image collection device, and determining illegal loading according to a classification result; and an operator terminal receiving information on whether objects are illegally loaded in the target cargo truck from the image processing device to display the received information. According to the present invention, the illegal loading enforcement is performed by using artificial intelligence learning, thereby providing an effect of increasing efficiency of enforcement work such as excluding closed cargo trucks not to be enforced.

Description

{System for enforcing bad and overloading vehicle to prevent falling object in cargo truck}

The present invention relates to a loading failure control system, and more particularly, to a loading failure control system for preventing falling objects in a freight vehicle.

Falling objects fall on the road due to poor loading of freight vehicles, resulting in many accidents and personal injury. However, since there is no quantitative standard for unloading of freight vehicles, the system to control it is in a situation where it is necessary to find the vehicle while checking the CCTV video 100% with the naked eye.

The safe loading of cargo is very closely related to the safe operation of vehicles on automobile-only roads, especially highways. To this end, it is essential that all cargoes, regardless of the type of cargo, must be securely fixed to the loading box so that they do not fall, and that the cargo does not exceed the loading box, and it is also necessary to comply with the prescribed loading rules.

Cargo load failure can cause fatal loss and harm not only to personnel and truck drivers involved in loading and unloading cargo, but also to other users, pedestrians, and the loaded cargo itself or vehicle. Monitoring must be blocked in advance for the safe operation of the vehicle.

Due to the recent industrial development and the spread of the Internet, the amount of movement of logistics (cargo) is increasing exponentially. From the shipper's point of view, overloading or speeding vehicles are being neglected to deliver a large amount of cargo in a short time as a way to solve the increasing volume of cargo, and due to this, the number of poorly loaded or speeding vehicles is continuously increasing. In particular, overloading damages the road, causes serious damage to people or vehicles around the body due to falling objects, and causes accidents in the 2nd and 3rd cars, which is more serious.

Cargo transportation means the safe movement of cargo to its destination. Cargo transportation includes air transportation by plane, port transportation through sea, or land transportation by freight car. In particular, for smooth maneuverability and door-to-door transport, the land transport must be finally carried out through a freight car, and the land transport can provide a consistent transport service without interlocking with other transport agencies. It is increasing significantly.

Accordingly, the number of accusations against poorly loaded freight vehicles and traffic accidents due to falling objects are increasing significantly, and as a result, damages to the lives and property of drivers are increasing day by day. In addition, social costs such as congestion of the highway due to the falling objects and treatment of falling objects are incurred, and accordingly, social interest in improving the crackdown and efficiency of poorly loaded vehicles is on the rise.

Article 39 (4) of the Road Traffic Act stipulates that'drivers of all vehicles shall take necessary measures to ensure that they are securely fixed, such as covering or tying up cargo so that the cargo does not fall while driving.'

Previously, the method of cracking down on poorly loaded vehicles was to install a camera at a business office, checkpoint, or toll gate, or operate a vehicle exclusively for enforcement with a camera capable of 360-degree rotation, and through this, photograph the loading status of the vehicle to crack down. .

This conventional method records the loading status of the vehicle through a camera as an image, and the crackdown is performed after a police officer visually checks it, and there is a problem that it is impossible to crack down for 24 hours.

In this way, when a person checks the collected images and determines that there is a risk of falling of the load, the existing crackdown on faulty loading is performing the task of prosecuting with reasonable reasons (bad binding, poor construction material cover, etc.). Therefore, due to the nature of the enforcement work, there is a problem in that all freight vehicles are inspected with the naked eye, resulting in human and economic losses.

Korean Patent Registration 10-1115589

The present invention has been conceived to solve the above problems, and an object of the present invention is to provide a loading defect control system capable of performing a crackdown by determining a loading defect of a freight vehicle using artificial intelligence learning.

The object of the present invention is not limited to the above-mentioned object, and other objects not mentioned will be clearly understood by those skilled in the art from the following description.

The present invention for achieving the above object relates to a loading defect control system, a photographing device for photographing the rear side of a freight vehicle passing through a road, an image collecting device for collecting an image photographed by the photographing device, and the image Includes a load classification engine for classifying the freight vehicle by performing a learning process using deep learning on the rear image of the freight vehicle collected by the collection device, and when an image of the target freight vehicle is input from the image collection device, the load is classified And an image processing device for classifying the target cargo vehicle through an engine and determining whether a load is defective according to the classification result, and an operator terminal receiving and displaying information on whether or not the target cargo vehicle is defective in loading from the image processing device.

The image processing device extracts a first feature region from the entire image including the loading unit from the rear image of the freight vehicle, and compares the extracted first feature region with a first table in which a list of classification names for each vehicle type is stored. Vehicles can be classified.

The image processing device extracts a second feature region for a door or a cover from a rear image of the freight vehicle, and compares the extracted second feature region with a second table in which a list of classification names for each vehicle type is stored. Can be classified.

As a result of classifying the target cargo vehicle, the image processing apparatus may determine that a cargo vehicle having a load and not a cover as a cargo vehicle carrying scrap metal and construction materials that cannot be bound is a defective loading.

As a result of classifying the target cargo vehicle, the image processing device loads vehicles that cannot be loaded, including passenger cars, dump trucks without cargo, ready-mixed concrete, aerial work vehicles, rescue special vehicles, cargo trucks without cargo, and trailers. Bad judgment can be excluded.

As a result of classifying the target freight vehicle, the image processing apparatus may exclude the determination of the unloading of a vehicle with a load and a cover.

As a result of classifying the target cargo vehicle, the image processing apparatus may determine that the vehicle is a crackdown vehicle that the inspector must directly check for the defective loading or the remaining vehicles that do not fall under the exclusion.

A sensor that generates a trigger signal when the freight vehicle passes, and may further include a trigger sensor for photographing a rear image of the freight vehicle by driving the photographing device by the trigger signal.

The loading failure control system may further include a WIM (Weigh-In-Motion) system for measuring the weight of a passing vehicle. At this time, when the WIM system detects the passage of the vehicle, the traffic signal is transmitted to the image collecting device, the image collecting device transmits the traffic signal received from the WIM system to the photographing device, and the photographing device passes Upon receiving the signal, you can take a picture of the rear of the vehicle.

The image processing apparatus may determine the presence or absence of a load of the target cargo vehicle using the weight information of the target cargo vehicle received from the WIM system.

According to the present invention, there is an effect that the efficiency of the enforcement work can be improved, such as excluding closed freight vehicles that do not need to be cracked, by performing the crackdown on defective loading using artificial intelligence learning.

In addition, according to the present invention, it is possible to find a vehicle that violates falling objects by searching for a cover for an open-type freight vehicle subject to a defective loading, and through this, the falling objects of the freight vehicle can be reduced. It can help to maintain a clean road and a safe road environment.

1 is a block diagram showing the configuration of a loading failure control system according to an embodiment of the present invention.
2 is a diagram showing an installation example of a loading failure control system according to an embodiment of the present invention.
3 and 4 show feature regions in a rear image of a freight vehicle according to an embodiment of the present invention.
5 is a diagram illustrating a classification of a freight vehicle in a tree-shaped structure through a load classification engine according to an embodiment of the present invention.
6 is a flowchart illustrating a process of determining a loading failure in a loading failure control system according to an embodiment of the present invention.
7 is a block diagram illustrating an example of a configuration of a system for controlling a loading failure according to an embodiment of the present invention.
8 is a chart showing result values according to classification results in the loading failure control system according to an embodiment of the present invention.
9 is a block diagram showing the configuration of a loading failure control system according to another embodiment of the present invention.
10 is a diagram showing an installation example of a loading failure control system according to another embodiment of the present invention.
11 to 15 show the classification of a freight vehicle in the present invention.
16 to 17 illustrate tables in which a list of classification names by vehicle type of a freight vehicle is stored in the present invention.

Advantages and features of the embodiments disclosed in the present specification, and a method of achieving them will become apparent with reference to the embodiments described later together with the accompanying drawings. However, the embodiments proposed in the present disclosure are not limited to the embodiments disclosed below, but may be implemented in various different forms, and only these embodiments are examples to those of ordinary skill in the art. It is only provided to fully inform the categories of the subjects.

The terms used in the present specification will be briefly described, and the disclosed embodiments will be described in detail.

Terms used in the present specification have selected general terms that are currently widely used as possible while taking the functions of the disclosed embodiments into consideration, but this may vary according to the intention or precedent of a technician engaged in a related field, the emergence of new technologies, and the like. In addition, in certain cases, there are terms arbitrarily selected by the applicant, and in this case, the meaning of the terms will be described in detail in the detailed description of the corresponding specification. Therefore, the terms used in the present disclosure should be defined based on the meaning of the term and the overall contents of the present specification, not a simple name of the term.

In the present specification, expressions in the singular include plural expressions unless the context clearly specifies that they are in the singular.

When a part of the specification is said to "include" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary. In addition, the term "unit" used in the specification refers to a hardware component such as software, FPGA, or ASIC, and "unit" performs certain roles. However, "unit" is not meant to be limited to software or hardware. The “unit” may be configured to be in an addressable storage medium or may be configured to reproduce one or more processors. Thus, as an example, "unit" refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, procedures, Includes subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, database, data structures, tables, arrays and variables. The functions provided within the components and "units" may be combined into a smaller number of components and "units" or may be further separated into additional components and "units".

In addition, in the description with reference to the accompanying drawings, the same reference numerals are assigned to the same components regardless of the reference numerals, and redundant descriptions thereof will be omitted. In describing the present invention, when it is determined that a detailed description of a related known technology may unnecessarily obscure the subject matter of the present invention, a detailed description thereof will be omitted.

The present invention is a defect loading control system that enables detection of defective loading of a freight vehicle through a single image on the rear using artificial intelligence imaging technology. The detection method detects a freight vehicle from a single image on the rear side and classifies the freight vehicle by feature point by learning the feature regions associated with defective loading in the detected freight vehicle. Can be distinguished.

In the present invention, in the case of an open-type freight vehicle, if there is a cover or strap, it is not defective, so it is possible to determine whether the cover or strap is through reinforcement of learning of the cover and strap in the loading area of the open-type freight vehicle.

The learning used in the present invention is a deep learning technology, and because it can achieve the same performance as the human eye judges with only a single image on the back, a number of existing cameras and additional sensors (laser scanner, ultrasound, etc.) are used. In terms of economy, accuracy, and maintainability, it can more effectively control misloading.

The present invention can be divided into step 1 and step 2. In the first step, cargo vehicles are classified by cargo loading type, and vehicles that are capable of being loaded and vehicles that are not capable of being loaded are classified, and vehicles that are not capable of being loaded are excluded from the subject of enforcement.

In the second step, reinforcement learning is performed on an area in which there is a possibility of defective loading among vehicles capable of poor loading to determine whether or not the loading is defective.

1 is a block diagram showing the configuration of a loading failure control system according to an embodiment of the present invention, Figure 2 is a diagram showing an installation example of the loading failure control system according to an embodiment of the present invention.

1 and 2, the loading defect control system of the present invention includes a photographing device 110, an image collection device 120, an image processing device 130, an operator terminal 140, and a trigger sensor 150. do.

The photographing device 110 serves to photograph the rear of the freight vehicle 10 passing through the road. For example, the photographing device 110 may be implemented as a camera.

The image collection device 120 serves to collect an image captured by the photographing device 110.

The image processing device 130 includes a load classification engine for classifying the freight vehicle by performing a learning process using deep learning on the rear image of the freight vehicle 10 collected by the image collection device 120, and , When the image of the target cargo vehicle is input from the image collection device 120, the target cargo vehicle is classified through the load classification engine, and whether or not the load is defective is determined according to the classification result.

The operator terminal 140 receives and displays information about whether or not the target freight vehicle is loaded from the image processing device 130. In the present invention, the operator terminal 140 includes a terminal capable of communicating through a wired or wireless communication network, and may be implemented as, for example, a PC, a desktop computer, a laptop computer, a tablet PC, and a smart phone.

The trigger sensor 150 is a sensor that generates a trigger signal when the freight vehicle passes, and the photographing device 110 is driven by the trigger signal to capture a rear image of the freight vehicle.

As shown in Fig. 2, the actual field configuration of the loading failure control system of the present invention is to install an image camera at the rear of the lane through which the freight vehicle passes, so that the rear image can be photographed when the freight vehicle 10 passes a designated point. do. By installing such a configuration on various roads, a vast amount of information on the rear of the freight vehicle is collected, and learning about the characteristic area designed in the present invention is proceeded. When the learning is completed, the load classification engine software of the present invention searches for a pre-learned feature region from the rear image of a passing vehicle, and at this time, selects the feature region with the highest similarity. In addition, a poorly loaded freight vehicle with a risk of falling can be detected by a combination of each of the feature regions.

The image processing apparatus 130 extracts a first feature region from the entire image including the loading part from the rear image of the freight vehicle 10, and compares the extracted first feature region with a first table in which a list of classification names for each vehicle type is stored. The target freight vehicle can be classified in such a way.

In addition, when there is an additional feature region, the image processing apparatus 130 extracts a second feature region for the door or cover part from the rear image of the freight vehicle 10, and uses the extracted second feature region as a list of classification names for each vehicle type. The target freight vehicle may be classified by comparing it with the stored second table.

As a result of classifying the target cargo vehicle, the image processing apparatus 130 may determine that a cargo vehicle having a load and not a cover as a cargo vehicle carrying scrap metal and construction materials that cannot be bound is a poor loading.

As a result of classifying the target cargo vehicle, the image processing device 130 is used for vehicles that cannot load cargo, including passenger cars, dump trucks without cargo, ready-mixed concrete, aerial work vehicles, rescue special vehicles, cargo trucks without cargo, and trailers. On the other hand, the judgment of poor loading can be excluded

In addition, as a result of classifying the target freight vehicle, the image processing apparatus 130 may exclude the determination of the load failure for the vehicle with the load and the cover.

As a result of classifying the target freight vehicle, the image processing apparatus 130 may determine that the remaining vehicles that are not subject to loading defects or exclusions are the enforcement vehicles that the inspector must directly check.

3 and 4 show feature regions in a rear image of a freight vehicle according to an embodiment of the present invention.

As shown in FIGS. 3 and 4, in the present invention, a prior classification is defined for each feature area for classification of a freight vehicle.

In the embodiments of FIGS. 3 and 4, the first feature areas 310 and 410 are areas including the entire vehicle area, and the second feature areas 320 and 420 indicate a rear door area or a rear circular area. I'm doing it.

The pre-classification for the freight vehicle of the present invention is defined in detail in FIGS. 11 to 15. 11 to 15 show the classification of a freight vehicle in the present invention.

In the present invention, a learning process using deep learning technology will be described as follows.

Step 1: In the unit image data, input the entire area for the target object and the coordinate values (left-top, right-bottom) in the image for a specific area, and predefined classification values as one data. For example, in FIGS. 11 to 15, a classification name may be input as a classification value.

Step 2: By inputting a large amount of data for each classification, the artificial intelligence software learns the number of different cases for each classification, and as a result of the cumulative learning, learning is performed at the point where the similarity of the feature regions no longer increases. It ends.

Step 3: The software developed as a result of learning searches for a previously learned object (target object) from the image when an untrained back image is input and calculates the classification value, coordinates, and similarity.

The above learning method can perform additional learning repeatedly whenever a new class is added. The software generated through such learning is hereinafter referred to as “loading classification engine”. When a freight vehicle is classified through a load classification engine, a tree-shaped structure is formed, and in the case of a specific freight vehicle, a specific node value is obtained.

5 is a diagram illustrating a classification of a freight vehicle in a tree-shaped structure through a load classification engine according to an embodiment of the present invention.

In the present invention, the photographing device 110 captures a rear image of a freight vehicle and transmits the image to the image collection device 120. The image collection device 120 transmits the collected images to the image processing device 130, and the image processing device 130 checks the classification result for the corresponding freight vehicle through the load classification engine.

For example, if the load classification result is classified as “DumpT” only, this vehicle is considered to be an uncovered dump truck. Because if there was a cover, the classification of the vehicle would have been judged as “DumpCov”. Based on this result, the image processing apparatus 130 determines that there is a risk of falling objects because the dump truck is not covered, and notifies the operator terminal 140 as a vehicle at risk of falling of a load.

11 to 15, in the case of an open freight vehicle, the cover and the tolerance are classified. Through this, it is possible to find a vehicle without a cover without a tolerance through the learning of the feature region for the cover and the feature region for the tolerance.

6 is a flowchart illustrating a process of determining a loading failure in a loading failure control system according to an embodiment of the present invention, and FIGS. 16 to 17 are tables in which a list of classification names for each vehicle type of a freight vehicle is stored in the present invention. Here, FIG. 16 shows Table 1, and FIG. 17 shows Table 2.

6, 16, and 17, when a vehicle enters from a lane and a rear image is collected and this image is input (S601), the image has the corresponding image through the load classification engine software implemented with an algorithm through learning. Classification for the feature region is executed (S603).

If the classification is successful (S605), each feature region is compared with the classification names in Table 1 to determine which classification name (Class Name) defined in the present invention, and if the corresponding classification name exists, the The classification name is stored (S607~S611).

If a plurality of feature regions exist in one image, the classification names of Table 2 are compared, and if the corresponding classification name exists, the classification name is stored (S613 to S617). In this way, the classification names defined in Table 1 and Table 2 are repeatedly searched according to the existence of the feature region.

By integrating the classifications stored through the above process, all the feature points of one image can be found (S619).

Then, the loading defect determination algorithm is executed, and it is determined whether or not the loading is defective according to the classification result (S621 to S629).

8 is a chart showing result values according to classification results in the loading failure control system according to an embodiment of the present invention.

As shown in FIG. 8, the present invention proposes an algorithm that derives three result values of “accusation”, “exclusion”, and “regulation” based on the classification name of this feature area.

Referring to FIG. 8, the result value "accusation" is a vehicle that can be accused immediately without intervention by the enforcement operator (S625), and the result value "excluded" is a case where a cover is placed on a closed vehicle or an open load that is impossible to load badly ( S627). In addition, the result value "regulation" is a case where the actual confirmation of the operator is necessary because it is not possible to automatically find whether the loading is defective (S629).

This result value derivation algorithm can be obtained by a combination of classification names that finds the characteristic structure that can cause falling objects of a freight vehicle, and it derives the result according to the following rules.

Result Value In the case of “accusation”, it is a case where the accusation condition is satisfied as a vehicle that is not covered with a separate cover even though there is a load of scrap metal and construction materials that cannot be bound, and the classification names are DumpT, CrabT, ReDumpT, etc.

In the case of “Excluding” the result value, there is only one feature area, but there is no need to crack down due to improper loading. The classification names at this time are PassCar, DumpTE, ReDumpTE, RmCon, CrabTE, CarcE, Bus, SkyS, SkyB, TowS, It can be TowB, CargoE, Trailer1E, Trailer2E, etc. In addition, when it is confirmed that the vehicle is inadequate to be loaded through two characteristic areas, for example, when there is a load in the dump, but it is normally loaded because there is a cover, when it is determined as a top vehicle or a tank lorry, but the load is actually loaded similarly. However, this is a case where the features such as BoxDoor and TankBack are combined and proved to be a top vehicle and a tank lorry. Classification names at this time include CrabT + CrabCov, BoxT + BoxDoor, TankT + TankBack, Cargo + CargoCov, DumpT + DumpCov, ReDumpT + ReDumpCov, etc.

In the case of “enforcement,” the result value is classified as a vehicle that is not subject to automatic prosecution or exclusion, and the classification name includes Cargo, BoxT, TankT, FishT, etc., and other unclassified vehicles are applicable. do.

Such an algorithm derives a result according to the above rules, and provides enforcement information to the operator terminal 140 accordingly.

7 is a block diagram illustrating an example of a configuration of a system for controlling a loading failure according to an embodiment of the present invention.

As shown in Figure 7 (a), in the present invention, the loading failure control system may be configured by separating the camera 710, the image collecting device 720, and the image processing device 730 into separate equipment.

Alternatively, as shown in FIG. 7 (b), a camera, an image sensor module, and an image processing module may be combined into one edge computer 810 by introducing the concept of an edge computer.

In addition, depending on the situation, the strobe 740, the lens 750, the trigger sensor 760, the operating computer 770, etc. may be combined and configured according to the situation.

9 is a block diagram showing a configuration of a loading failure control system according to another embodiment of the present invention, and FIG. 10 is a diagram illustrating an installation example of a loading failure control system according to another embodiment of the present invention.

9 and 10, the loading defect control system according to another embodiment of the present invention includes a photographing device 110, an image collection device 120, an image processing device 130, an operator terminal 140, and a WIM ( Weigh-In-Motion) system 900.

The photographing device 110 serves to photograph the rear of the freight vehicle 10 passing through the road. For example, the photographing device 110 may be implemented as a camera.

The image collection device 120 serves to collect an image captured by the photographing device 110.

The image processing device 130 includes a load classification engine for classifying the freight vehicle by performing a learning process using deep learning on the rear image of the freight vehicle 10 collected by the image collection device 120, and , When the image of the target cargo vehicle is input from the image collection device 120, the target cargo vehicle is classified through the load classification engine, and whether or not the load is defective is determined according to the classification result.

The operator terminal 140 receives and displays information about whether or not the target freight vehicle is loaded from the image processing device 130. In the present invention, the operator terminal 140 includes a terminal capable of communicating through a wired or wireless communication network, and may be implemented as, for example, a PC, a desktop computer, a laptop computer, a tablet PC, and a smart phone.

The WIM system 900 is a system for measuring the weight of a passing vehicle.

The WIM system 900 includes a WIM controller 910 for controlling the system, and an axle pad 920 for detecting an entering vehicle and measuring weight.

When the WIM system 900 detects the traffic of the vehicle, the traffic signal is transmitted to the image collecting device 120, and the image collecting device 120 captures the traffic signal received from the WIM system 900 by the photographing device 110. And the photographing device 110 photographs the rear of the vehicle when receiving a traffic signal.

The image processing device 130 determines the presence or absence of a load of the target freight vehicle using the weight information of the target freight vehicle received from the WIM system 900.

As described above, in the present invention, it is possible to more effectively implement a loading failure control system through linkage with the WIM system.

In the case of a video image, a change occurs according to the weather and a shadow, and thus, in the case of such a freight vehicle image, an error that confuses the tolerance, which means a freight vehicle without a load, may occur. In order to reduce these errors, the system can be configured by combining with the WIM system.

That is, the WIM system can obtain the total weight of the freight vehicle, and can convert the vehicle's empty vehicle weight into a database through financial information on the classified vehicles. Therefore, after subtracting the empty vehicle weight from the gross weight, it becomes the loaded weight of the vehicle, and it is possible to check whether the vehicle is empty.

Looking at the operating procedure of the loading defect control system linked to the WIM system, the image collection device 120 receives a signal through the WIM system 900 to check the timing of collecting the rear image of the vehicle, and this is the photographing device 110. Deliver the shooting order to the person.

In addition, the image collection device 120 transmits the rear image and the WIM weight information to the image processing device 130 after the vehicle passes and the weight information is obtained from the WIM system 900.

After classifying the vehicle through the load classification engine, the image processing apparatus 130 may know whether or not a tolerance is made using WIM weight information. For example, when the loading section is deep, such as a dump, a tong truck, etc., or when there is an installation device on the loading section, or in the case of a general freight vehicle in bad weather conditions such as snow or rain, WIM weight information can be used to improve the accuracy of the tolerance. have.

In addition to the purpose of determining tolerance, weight information of the WIM system can also be detected for uneven loading. It is very dangerous if the weight is unevenly loaded on the left and right when loading the weight on the loading part among the vehicles subject to the video-type poor loading, because it can cause an accident when driving on a corner road. The loading defect control system of the present invention determines whether there is uneven loading based on the weight of each wheel transmitted from the WIM system, so that the enforcement work can be performed more efficiently. An example of the enforcement standards at this time is as follows.

Gross weight (including loads): GW (Gross Weight)

Sum of left wheel weight: LGW (Left GW)

Sum of right wheel weight: RGW (Right GW)

Unbiased loading conditional expression: | RGW-LGW | > GW * 30%

In other words, if the deviation of the sum of the left and right wheel weights exceeds 30% of the total weight of the vehicle, it can be judged as unbiased loading and prosecuted.

In this way, it is possible to more accurately control the load failure by using the load weight information of the WIM system and whether or not the open-type freight vehicle is uncovered.

Although the present invention has been described using several preferred embodiments, these embodiments are illustrative and not limiting. Those of ordinary skill in the art to which the present invention pertains will understand that various changes and modifications can be made without departing from the spirit of the present invention and the scope of the rights set forth in the appended claims.

110 Recording device 120 Image collecting device
130 Image processing unit 140 Operator terminal
150 trigger sensor

Claims (10)

A photographing device for photographing the rear side of a freight vehicle passing through the road;
An image collection device for collecting an image photographed by the photographing device;
It includes a load classification engine for classifying the freight vehicle by performing a learning process using deep learning on the rear image of the freight vehicle collected by the image collection device, and when an image of the target freight vehicle is input from the image collection device, the An image processing device for classifying the target cargo vehicle through a load classification engine and determining whether or not the load is defective according to the classification result; And
Including an operator terminal for receiving and displaying information on whether or not the loading of the target freight vehicle is defective from the image processing device,
The image processing device extracts a first feature region from the entire image including the loading unit from the rear image of the freight vehicle, and compares the extracted first feature region with a first table in which a list of classification names for each vehicle type is stored. Classify vehicles,
The image processing device extracts a second feature region for a door or a cover from a rear image of the freight vehicle, and compares the extracted second feature region with a second table in which a list of classification names for each vehicle type is stored. To classify,
As a result of the comparison, the image processing device is a “accusation” in case of a poorly loaded vehicle that can be directly accused without intervention by the enforcement operator, based on the classification name of the vehicle as a result of the comparison. In the case of a vehicle in which the enforcement operator must directly check whether “excluded” or the loading is defective, the result value of “regulation” is output,
The image processing device is a vehicle that carries scrap metal and construction materials that cannot be bound and satisfies the accusation condition as a vehicle that is not covered with a separate cover even if there is a load. Knuckle crane truck), ReDumpT (remodeled dump truck with cargo), print "Complaint",
When there is only one characteristic area and there is no need to regulate because of improper loading, PassCar (passenger car), DumpTE (dump truck without cargo), ReDumpTE (modified dump truck without cargo), RmCon (remicon), CrabTE (cargo) Knuckle crane truck with no cargo), CarcE (car for transporting vehicles without cargo), Bus (bus), SkyS (high place work car (small)), SkyB (high place work car (large)), TowS (recovery special car (small) )), TowB (rescue type special vehicle (large)), CargoE (cargo truck without cargo), Trailer1E (container type trailer without cargo), Trailer2E (low-bed trailer without cargo), Print "exclude",
It is a case where there are two characteristic areas and there is no need for crackdown because of improper loading. CrabT (knuckle crane truck with cargo) + CrabCov (knuckle crane truck cover), BoxT (wing body vehicle, refrigerated/frozen vehicle) + BoxDoor ( Wing body vehicle, rear door of refrigerated/frozen vehicle), TankT (tank lorry) + TankBack (rear of tank lorry), Cargo (cargo truck with cargo) + CargoCov (cargo truck cover), DumpT (dump truck with cargo) + DumpCov ( Dump truck cover), ReDumpT (modified dump truck with cargo) + ReDumpCov (modified dump truck cover), output "Excluded",
Classification names including Cargo (cargo truck with cargo), BoxT (wing body vehicle, refrigerated/frozen vehicle), TankT (tankrower), and FishT (live fish transport vehicle) as the vehicle not subject to the above "accusation" and "excluded" In the case of, "continue"
Further comprising a WIM (Weigh-In-Motion) system for measuring the weight of the passing vehicle,
When the WIM system detects the passage of the vehicle, the traffic signal is transmitted to the image collecting device, the image collecting device transmits the traffic signal received from the WIM system to the photographing device, and the photographing device is a traffic signal When receiving, shoots the rear of the vehicle,
The image processing device determines the presence or absence of a load of the target cargo vehicle using the weight information of the target cargo vehicle received from the WIM system, and refers to the database in which the empty vehicle weight of the cargo vehicle is stored, and the total weight and tolerance of the target cargo vehicle Checking the tolerance of the target freight vehicle by comparing the weight,
When the deviation of the sum of the left wheel weight and the right wheel weight of the specific target freight vehicle exceeds 30% of the total vehicle weight through the WIM system, the image processing device determines that it is unevenly loaded and outputs a "accusation" result value. Loading defect control system characterized by.
delete delete delete delete delete delete The method according to claim 1,
A sensor for generating a trigger signal when the freight vehicle passes, and further comprising a trigger sensor for photographing a rear image of the freight vehicle by driving the photographing device by the trigger signal.
delete delete

Images