阅读说明：本技术 一种仓库货物周边异常事件报警方法及系统 (Warehouse goods peripheral abnormal event alarm method and system ) 是由 丁柏宇辉 王韬 方睿 范子鑫 于 2021-07-06 设计创作，主要内容包括：本发明公开一种仓库货物周边异常事件报警方法及系统,检测服务器接收采集设备采集的货物图片,检测服务器通过目标检测算法对货物图片检测,输出检测结果；逻辑服务器运行仓库信息管理系统,逻辑服务器接收检测服务器的检测结果,输出货物周边异常事件,监控终端接收货物周边异常事件。报警方法系统最大限度的利用既有的监控摄像头,减少了感知设备和人力的消耗,提高货物巡检效率和效果。(The invention discloses a warehouse cargo peripheral abnormal event alarm method and system.A detection server receives a cargo picture acquired by acquisition equipment, detects the cargo picture through a target detection algorithm and outputs a detection result; the logic server operates the warehouse information management system, receives the detection result of the detection server, outputs the abnormal events around the goods, and the monitoring terminal receives the abnormal events around the goods. The alarm method system utilizes the existing monitoring camera to the maximum extent, reduces consumption of sensing equipment and manpower, and improves cargo inspection efficiency and effect.)

1. A warehouse goods peripheral abnormal event alarm method is characterized in that: be applied to warehouse goods peripheral abnormal event alarm system, alarm system includes: the system comprises acquisition equipment, a detection server, a logic server and a monitoring terminal; the alarm method comprises the following steps:

the method comprises the steps that a detection server receives goods pictures collected by collection equipment, wherein the collection equipment comprises a camera group arranged on a warehouse stand column, the camera group comprises a vertical camera and a plane camera, and the goods pictures comprise pictures shot by the vertical camera and pictures shot by the plane camera;

the detection server detects the goods picture through a target detection algorithm and outputs a detection result;

wherein the target detection algorithm comprises: at least one of a safety helmet detection algorithm, a reflective vest detection algorithm, a forklift detection algorithm, and a cargo detection algorithm;

the logic server operates the warehouse information management system, receives the detection result of the detection server, outputs the abnormal events around the goods, and the monitoring terminal receives the abnormal events around the goods.

2. The alarm method of claim 1: the method is characterized in that: the abnormal events comprise that the safety helmet is not worn by the operator, the reflective vest is not worn by the operator, the forklift is not in the warehouse entry and exit bill time period, and the quantity of the goods is inconsistent.

3. The alarm method of claim 2: the method is characterized in that: the safety helmet detection algorithm, the reflective vest detection algorithm, the forklift detection algorithm and the cargo detection algorithm all adopt the Yolo V5 algorithm.

4. The alarm method of claim 3: the method is characterized in that: the detection result comprises the following steps: if the safety helmet detection algorithm detects that the operator in the cargo picture wears the safety helmet, the abnormal event around the cargo is determined to be safe, and if the operator in the cargo picture does not wear the safety helmet, the abnormal event around the cargo is determined to be abnormal.

5. The alarm method of claim 4: the method is characterized in that: the detection result comprises the following steps: if the reflective vest detection algorithm detects that the operator in the goods picture wears the reflective vest, the abnormal event around the goods is determined to be safe, and if the operator in the goods picture does not wear the reflective vest, the abnormal event around the goods is determined to be abnormal.

6. The alarm method of claim 5: the method is characterized in that: the detection result comprises the following steps: the fork truck detection algorithm detects the fork truck in the warehouse and records the time that the fork truck appears in the warehouse location.

7. The alarm method of claim 6: the method is characterized in that: and the logic server receives the time when the forklift appears in the warehouse location, compares the time with the time period of the warehouse entering and exiting order sent by the warehouse information management system, and determines that the abnormal event around the goods is abnormal if the time is not in the time period.

8. The alarm method of claim 7: the method is characterized in that: the detection result comprises the following steps: and detecting the quantity of the goods in each storage position through a goods detection algorithm.

9. The alarm method of claim 8: the method is characterized in that: the goods detection algorithm detects a single tray and a stack on the tray, records coordinates and center coordinates of four corners of a rectangle detected by each goods and the tray, calculates the number of stacked layers of the goods and the number of the goods in each row, and calculates the number of the goods in each storage position.

10. The alarm method of claim 9: the method is characterized in that: and the logic server receives the quantity of the goods of each storage position, compares the quantity of the goods with the quantity of the goods recorded in the warehouse bill in the warehouse information management system, and determines that the abnormal events around the goods are abnormal if the quantities are inconsistent.

11. The warehouse goods peripheral abnormal event alarm system is characterized in that: the alarm system comprises acquisition equipment, a detection server, a logic server and a monitoring terminal, wherein the acquisition equipment and the detection server are positioned in an intranet, the acquisition equipment and the detection server are connected in a wired mode for communication, the logic server and the monitoring terminal are positioned in an extranet, the logic server and the monitoring terminal are connected in a wired mode for communication, the detection server and the logic server are connected in a wired mode for communication, and the detection server, the logic server and the monitoring terminal are used for executing the alarm method according to one of claims 1-10.

Technical Field

The invention relates to the technical field of warehouse information management systems, in particular to a warehouse goods peripheral abnormal event alarm method and system based on a target detection algorithm.

Background

The financing of the movable property is a business of bank credit. The bank has ownership and security rights to the pledge's footage. The bank has the right to the mobile asset, but the mobile asset is stored by the bank, and the bank needs to pay certain storage cost of the warehouse while the bank takes the mobile asset. The warehouse can cause the damage or the extinction of the movable property due to poor storage, and the warehouse needs to take damage compensation responsibility to a bank, so the warehouse needs a set of management system for the movable property risk.

At present, the warehouse for storing the mobile assets has problems in management, for example, the goods in the warehouse still adopt a manual supervision mode, although strict warehouse management regulations and warehouse-in and warehouse-out operation flows are established, it is difficult to monitor whether an operator follows execution and the execution strength, and risks are brought to the warehouse. Even the warehouse is internally provided with the monitoring camera, the management personnel are also required to manually watch the videos of the abnormal events around the goods in the monitoring camera, so that the labor cost is high, and the management personnel can also miss the videos of the abnormal movement of the goods at times.

In order to guarantee the safety of goods among the prior art, some warehouse information management systems increase perception equipment through receiving the inside of installing at the goods or on the surface, for example the response information of equipment such as induction type label, acceleration sensor, but to every goods subsides RFID chip to go out the warehouse entry action to every goods and carry out the discernment management, will consume a large amount of manpowers and financial resources to the transformation in warehouse, be unfavorable for the construction cost and the complexity requirement in warehouse.

In addition, because the goods in the large warehouse are very many, although monitoring cameras are also very many, managers are difficult to concentrate on, and monitor the goods abnormal movement or the loss of goods in time, further cause the bank to be difficult to supervise the goods.

Disclosure of Invention

Aiming at the situation, the invention provides a method for detecting safety helmets, reflective vests, forklifts and goods by a target detection algorithm to monitor the operation specification of operators and give an alarm in real time when the goods are illegally moved. The alarming method is characterized in that monitoring cameras at different positions in a warehouse are used for polling different warehouse positions, the number of goods in the warehouse positions detected through a target detection algorithm is compared with the number of the goods recorded in a warehouse receipt, and alarming processing is carried out on the warehouse positions with inconsistent numbers. The alarm system does not need to add extra sensing equipment to the goods, and the existing monitoring camera is utilized to carry out target detection, so that the construction cost of the warehouse is reduced, the complexity of the system is reduced, self-monitoring of the warehouse is realized, and the goods inspection efficiency and effect are improved.

In order to improve the degree of executing the operation flow of the operator and reduce the risk of goods loss, the invention also provides a warehouse goods peripheral abnormal event alarm method, which is applied to a warehouse goods peripheral abnormal event alarm system, wherein the alarm system comprises: the system comprises acquisition equipment, a detection server, a logic server and a monitoring terminal; the alarm method comprises the following steps: the detection server receives the goods picture acquired by the acquisition equipment; the acquisition equipment comprises a camera group arranged on a warehouse upright post, the camera group comprises a vertical camera and a plane camera, and the goods picture comprises a picture shot by the vertical camera and a picture shot by the plane camera; the detection server detects the goods picture through a target detection algorithm and outputs a detection result; wherein, the target detection algorithm comprises: at least one of a safety helmet detection algorithm, a reflective vest detection algorithm, a forklift detection algorithm, and a cargo detection algorithm; the logic server operates the warehouse information management system, receives the detection result of the detection server, outputs the abnormal events around the goods, and the monitoring terminal receives the abnormal events around the goods.

Further, the abnormal events comprise that the safety helmet is not worn by an operator, the reflective vest is not worn by the operator, the forklift is not in the warehouse entry and exit bill time period, and the quantity of the goods is inconsistent.

Further, a safety helmet detection algorithm, a reflective vest detection algorithm, a forklift detection algorithm and a cargo detection algorithm all adopt a Yolo V5 algorithm.

Further, the detection result comprises: if the safety helmet detection algorithm detects that the operator in the cargo picture wears the safety helmet, the abnormal event around the cargo is determined to be safe, and if the operator in the cargo picture does not wear the safety helmet, the abnormal event around the cargo is determined to be abnormal.

Further, the detection result comprises: if the reflective vest detection algorithm detects that the operator in the goods picture wears the reflective vest, the abnormal event around the goods is determined to be safe, and if the operator in the goods picture does not wear the reflective vest, the abnormal event around the goods is determined to be abnormal.

Further, the detection result comprises: the fork truck detection algorithm detects the fork truck in the warehouse and records the time that the fork truck appears in the warehouse location.

Further, the logic server receives the time when the forklift is in the warehouse location, compares the time with the time period of the warehouse entering and exiting bill command sent by the warehouse information management system, and determines that the abnormal event around the goods is abnormal if the time is not in the time period.

Further, the detection result comprises: the cargo detection algorithm detects the quantity of cargo per bin.

Further, the goods detection algorithm detects a single tray and a stack on the tray, records coordinates and center coordinates of four corners of a rectangle detected by each goods and the tray, calculates the number of stacked layers of the goods and the number of the goods in each column, and calculates the number of the goods in each storage position.

Further, the logic server receives the quantity of the goods in each storage position, compares the quantity of the goods with the quantity of the goods recorded in the warehouse bill in the warehouse information management system, and sends an abnormal event alarm to the monitoring terminal through the logic server if the quantity of the goods is inconsistent.

The alarm system comprises acquisition equipment, a detection server, a logic server and a monitoring terminal, wherein the acquisition equipment and the detection server are positioned in an intranet, the acquisition equipment and the detection server are connected in a wired mode for communication, the logic server and the monitoring terminal are positioned in an extranet, the logic server and the monitoring terminal are connected in a wired mode for communication, the detection server and the logic server are connected in a wired mode for communication, and the detection server, the logic server and the monitoring terminal are used for executing the alarm method.

Compared with the prior art, the invention has the following beneficial effects:

the alarm system for the abnormal events around the goods in the warehouse can detect whether an operator wears reflective clothes and a safety helmet and whether a forklift works abnormally, and can also judge whether the quantity of the goods is accurate and determine the abnormal events around the goods. And comparing the number of the cargos with the number of the cargos recorded in a warehouse bill in the warehouse information management system in real time, and judging whether the number of the cargos is consistent. The alarming system for the abnormal events around the goods in the warehouse utilizes the existing monitoring cameras to the maximum extent, reduces consumption of sensing equipment and manpower, and improves goods inspection efficiency and effect.

Drawings

FIG. 1 shows a diagram of a cargo alarm system;

FIG. 2 shows a flow chart of a cargo alarm method;

FIG. 3 shows a flow chart of an object detection algorithm;

FIG. 4 is a picture taken by a facade camera and a plan camera;

FIG. 5 shows a graph of the effectiveness of the cargo detection algorithm;

fig. 6 shows a detection effect diagram of the safety helmet.

Reference numerals: 10. an alarm system; 11. collecting equipment; 12. detecting a server; 13. a logical server; 14. a monitoring terminal; 11a, a vertical camera; 11b, a plane camera; 20. goods; 21. a column; 22. a tray; 23. a helmet; 24. a reflective vest; 25. a forklift; a1, side area; a2, planar area; in, inner net; wn, outer net; and S, operating personnel.

Detailed Description

In order to make the technical solutions of the present invention better understood by those skilled in the art, the technical solutions in the embodiments of the present invention will be clearly described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Fig. 1 shows an alarm system 10 for abnormal events around warehouse goods, where the alarm system 10 includes a collection device 11, a detection server 12, a logic server 13, and a monitoring terminal 14. The acquisition device 11 and the detection server 12 are located in an internal network in, the acquisition device 11 and the detection server 12 communicate with each other in a wired or wireless manner, the logic server 13 and the monitoring terminal 14 are located in an external network wn, the logic server 13 and the monitoring terminal 14 communicate with each other in a wired manner, and the detection server 12 and the logic server 13 communicate with each other in a wired manner from the internal network in to the external network wn. Wherein the extranet wn is connected to the internet so that the warehouse manager views the abnormal events around the goods in the warehouse through the internet.

Fig. 2 shows a warehouse goods periphery abnormal event alarm method, which is applied to a warehouse goods periphery abnormal event alarm system 10, wherein the alarm system 10 includes: the system comprises acquisition equipment 11, a detection server 12, a logic server 13 and a monitoring terminal 14. The alarm method comprises the following steps: the detection server 12 receives the goods pictures acquired by the acquisition equipment 11, wherein the acquisition equipment 11 comprises a camera group arranged on the warehouse upright post 21, the camera group comprises a vertical camera 11a and a plane camera 11b, and the goods pictures comprise pictures shot by the vertical camera 11a and pictures shot by the plane camera 11 b; the detection server 12 detects the cargo picture through a target detection algorithm and outputs a detection result, wherein the target detection algorithm includes: at least one of a safety helmet detection algorithm, a reflective vest detection algorithm, a forklift detection algorithm and a cargo detection algorithm, wherein the reflective vest detection algorithm, the safety helmet detection algorithm, the forklift detection algorithm and the cargo detection algorithm all adopt a Yolo V5 algorithm; the logic server 13 operates the warehouse information management system, the logic server 13 receives the detection result output by the detection server 12, the warehouse information management system compares the warehouse information with the detection result to determine the abnormal event around the goods, and the monitoring terminal 14 receives the abnormal event around the goods.

The abnormal events comprise that the operator S does not wear the safety helmet 23, the operator S does not wear the reflective vest 24, the forklift 25 does not appear in the warehouse entry and exit bill time period, and the quantity of the goods 20 is inconsistent.

The warehouse is internally provided with upright posts 21 for supporting, a warehouse position is formed between the upright posts 21, and the elevation camera 11a and the plane camera 11b are both arranged at the top ends of the upright posts 21. The elevation camera 11a is used for shooting the side area of the goods 20, the plane camera 11b mainly shoots the plane area of the goods 20, and the goods pictures comprise pictures shot by the elevation camera 11a and pictures shot by the plane camera 11 b. Each library position is guaranteed to be shot by the elevation camera 11a and the plane camera 11 b.

In order to prevent the cargo 20 from being moved illegally, the operator S around the cargo 20 wears the forklift 25, and the cargo 20 is subjected to target detection, thereby identifying an abnormal event around the cargo.

To determine the anomalous event around the cargo, the detection server 12 runs a target detection algorithm. The detection server 12 detects the cargo picture in the video through a target detection algorithm, outputs a detection result, and sends the detection result to the logic server 13.

In order to detect the degree of the operator' S work specification, the safety helmet detection algorithm is used to perform target detection on whether the operator S around the cargo 20 in the cargo picture wears the safety helmet 23. If the detection server 12 detects that the operator S in the picture of the cargo wears the safety helmet 23, it is determined that the abnormal event around the cargo is safe, and if it is detected that the operator S in the picture of the cargo does not wear the safety helmet 23, it is determined that the abnormal event around the cargo is abnormal.

In order to detect the degree of the operator' S work specification, the reflective vest detection algorithm is used to perform target detection on whether the operator S around the cargo 20 in the cargo picture wears the reflective vest 24. If the detection server 12 detects that the operator S in the picture of the goods wears the reflective vest 24, it is determined that the abnormal event around the goods is safe, and if the operator S in the picture of the goods does not wear the reflective vest 24, it is determined that the abnormal event around the goods is abnormal.

In order to ensure that the forklift is in a certain event section, the forklift is prevented from entering a warehouse in an abnormal time period, and the risk that goods are removed is prevented. The fork truck detection algorithm detects the fork truck 25 in the warehouse and records the time that the fork truck 25 is present at the warehouse location. The logic server 13 receives the time when the forklift 25 appears in the warehouse location, compares the time with the time period of the warehouse entering and exiting bill command sent by the warehouse information management system, and determines that the abnormal event around the goods is abnormal if the time is not in the time period.

The goods detection algorithm detects the quantity of goods 20 per bin. The goods detection algorithm detects a single pallet 22 and a stack on the pallet 22, records the coordinates and center coordinates of the four corners of the rectangle detected by each goods 20 and pallet 22, calculates the number of stacked layers of the goods 20 and the number of goods 20 in each column, and calculates the number of goods 20 in each bay.

The logic server 13 operates the warehouse information management system, the logic server 13 receives the detection result of the detection server 12, the logic server 13 compares the warehouse slip information in the warehouse information management system with the detection result, and outputs the abnormal event around the goods, and the monitoring terminal 14 receives the abnormal event around the goods. The information of the warehouse entry and exit is pre-entered information of the warehouse entry and exit, and the information of the warehouse entry and exit comprises the number of the goods 20 in the warehouse entry and exit.

The logic server 13 compares the quantity of the goods 20 in each storage position with the quantity of the goods 20 in and out of the corresponding warehouse, if the quantities are inconsistent, the abnormal event around the goods is determined to be abnormal, and an alarm is sent to the monitoring terminal.

FIG. 3 shows a schematic flow diagram of an object detection algorithm, wherein the object detection algorithm comprises the following steps:

s101, acquiring a cargo picture;

s102, preprocessing a goods picture;

s103, detecting the dressing of an operator S in the goods picture, the forklift 25 and the goods 20.

S101, the step of acquiring the goods picture comprises the following steps: the collecting device 11 collects videos of goods 20 on the warehouse location, sends the collected videos to the detection server 12, and intercepts goods pictures of corresponding frames through Rtsp/Rtmp video streams sent by the camera. To improve the effect of the object detection algorithm, as shown in fig. 4, the left half is a picture taken by the elevation camera 11a, and the right half is a picture taken by the plane camera 11 b. In the picture taken by the elevation camera 11a, the side area a1 of the cargo 20 is larger than the plane area a2 of the cargo 20. In the picture taken by the plane camera 11b, the plane area a2 of the cargo 20 is larger than the side area a1 of the cargo 20.

S102, the step of preprocessing the goods picture comprises the following steps: to reduce the impact of dim light in the exposure warehouse environment on the performance of the target detection algorithm and to alleviate the requirement on the generalization capability of the target detection algorithm, the detection server 12 processes the brightness, contrast, and sharpness of the picture of the good.

S103, detecting the clothes of the operator S, the forklift 25 and the goods 2 in the picture of the detected goods by using a target detection algorithm.

Fig. 5 shows a diagram of the effect of the cargo detection algorithm of the present invention. The pictures taken by the elevation camera 11a and the pictures taken by the plane camera 11b are detected by a goods detection algorithm, and a single tray 22 and a stack on the tray 22 are detected.

Based on the detected coordinates of the four corners of the rectangle of each cargo 20 and pallet 22 and the coordinates of the center, the number of stacked layers of the cargo 20 and the number of the cargo 20 in each column are calculated, and the number of the cargo 20 in each stock space is calculated.

The actual quantity is the quantity of the goods 20 in the warehouse location calculated by the goods detection algorithm, the system quantity is the quantity of the goods 20 in and out of the warehouse by inquiring the warehouse bill, if the actual quantity is consistent with the system quantity, the detected quantity is correct, and if not, an alarm is given.

Fig. 6 shows a safety helmet detection algorithm effect graph. And detecting the pictures shot by the vertical camera 11a by using a reflective vest detection algorithm, a safety helmet detection algorithm and a forklift detection algorithm. The box in fig. 6 is the safety helmet detection algorithm detecting that the operator S is not wearing the safety helmet 23, and the reflective vest detection algorithm detecting that the operator S is wearing the reflective vest 24.

In order to ensure the real-time performance of the alarm system 10 for the abnormal events around the goods in the warehouse, the alarm system 10 patrols one round every minute, and the patrolling one round is to detect the pictures of the goods shot by a plurality of groups of monitoring cameras once. If the detected quantity of the goods 20 in the storage location is correct, the next storage location is checked. If the quantity of the goods 20 in the warehouse location is inconsistent with the quantity of the goods 20 in and out of the warehouse in the warehouse list in the inspection process, determining that the abnormal event around the goods is abnormal, and alarming.

According to the technical scheme, the picture to be detected sent by the acquisition equipment 11 is detected through the detection server 12, the picture to be detected is automatically identified by using a target detection algorithm to obtain a detection result, whether alarm information is sent to the monitoring terminal 14 or not is determined according to the detection result, the cost of labor overhead of a warehouse site is reduced, the problem of missed detection of warehouse site management is solved, the cost of manual operation is greatly reduced, and the method and the system have good economical efficiency and practicability.

While the foregoing is directed to the preferred embodiment of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.