Product conveying system and method based on machine vision

Abstract

The invention relates to a product conveying system and method based on machine vision, and relates to the technical field of logistics transportation control, and the product conveying system comprises a visual monitoring module which is used for recognizing and analyzing a product identification code to obtain a product target position, the running state of the product is tracked and traced, and the running state of each section of conveyor in the article transportation module is controlled; the article transportation module is used for conveying the product according to the product target position; the remote control module is used for connecting and controlling the visual monitoring module and the article transportation module according to an equipment access address, a wireless network account and a wireless network password; the single electric roller shafts are uniformly distributed in the transportation structure, so that the transportation efficiency of the material box is favorably improved; the operation states of the material box and the electric roller shaft are dynamically monitored in real time through a visual scheme, and the transportation efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of logistics transportation control technology, and in particular to a product conveying system and method based on machine vision. Background Technology

[0002] When a logistics transportation system is long enough, multiple transportation structures need to be spliced ​​together to achieve the required transportation length. The PLC wiring harness mainly connects to sensors and electric rollers to control their operation. This part of the work is the responsibility of electrical engineers, while the remaining frame parts are often the responsibility of mechanical engineers. Both are assembled according to customer requirements after being transported to the construction site. This results in situations where, for example, there may be one electric roller in every 10 meters of transportation structure, or there may be two or even more electric rollers in every 10 meters of transportation structure. Thus, the number of electric rollers in the entire logistics transportation system is uneven.

[0003] Based on this, the sensor's operating logic is as follows: the electric roller rotates to convey the material box on the transport structure. The material box passes through the first sensor and exits from the tail sensor before the next material box can re-enter the area between the first and tail sensors; this process is repeated to realize the flow of multiple material boxes on the transport system.

[0004] Application number CN202511122037.6 discloses a machine vision-based departing baggage check-in control system, which can optimize and simplify the baggage check-in process, improve check-in efficiency, optimize airport operations, and improve passenger experience. The technical solution is as follows: The system includes: a user interface module for providing airport staff with an operation and monitoring interface; a vision acquisition module for real-time monitoring of the baggage check-in status and capturing images and videos of the baggage during transportation; a vision processing module for collecting and processing data from the vision acquisition module and the control module, identifying the baggage status through image processing and algorithm analysis, and tracking any abnormalities that may occur during baggage transportation; and a control module for controlling the operation and stop of the corresponding belt conveyor based on the output of the vision processing module.

[0005] The existing technical solutions mentioned above have the following drawbacks: 1. Traditional PLC control systems can only realize the mechanical transportation of material boxes on the transportation system. If the sensor detects that the material box is too heavy or stuck at a corner and cannot be moved, it can only alarm and call for manual operation, which is too limited; 2. The existing logistics transportation system has an uneven layout, resulting in too many or too few material boxes in some areas, and low transportation efficiency. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the objective of this invention is achieved through the following technical solution: A product conveying system based on machine vision; comprising: The visual monitoring module is used to identify and parse product identification codes, obtain the target location of the product, track and trace the operating status of the product, and control the operating status of each section of the conveyor in the goods transportation module. The item transportation module is used to transport the product according to the target location of the product; The remote control module is used to connect to and control the visual monitoring module and the goods transportation module based on the device access address, Wi-Fi account and Wi-Fi password.

[0007] By adopting the above technical solutions, the visual monitoring module uses deep learning algorithms (such as YOLOv5+CRNN) to identify product identification codes and parse target locations in real time, while simultaneously combining Kalman filtering and multi-target tracking algorithms to continuously monitor product status; the goods transportation module transports products based on A* path planning and PID control algorithms, and embeds a real-time anomaly detection layer (dynamically calculating displacement thresholds and trajectory direction tolerances)—when displacement changes abruptly (such as GPS drift exceeding the threshold), the power is immediately frozen, or continuous directional anomalies trigger a 50% speed reduction and redundant navigation switching; the remote control module uses TLS encrypted tunnels and MQTT protocols to achieve secure interconnection of devices, ultimately forming a closed loop of "identification-transportation-monitoring-control"; the fault tolerance is greatly improved through multi-module intelligent collaboration, and the adaptive learning mechanism (such as window threshold rolling updates and multi-sensor voting decision-making) enables the system to maintain a high and stable delivery rate even in complex environments.

[0008] The present invention is further configured such that the visual monitoring module includes: The monitoring and acquisition unit is used to monitor and acquire the product's operating status to obtain the product's video stream. The data processing unit is used to perform frame decomposition on the product delivery video stream to obtain several product delivery images; The template recognition unit is used to match and filter the product delivery images one by one according to the preset identification template and the recognition clarity threshold to obtain a number of product identification images. The identifier parsing unit is used to parse the product identifier image and aggregate the same product identifier code to obtain the target location; The equipment monitoring unit is used to monitor the operating status of each section of the conveyor in the goods transportation module and obtain the equipment operating parameters; The tracking and tracing unit is used to track and locate the real-time position of several products and trace their operation according to preset outbound constraints, and to correct the operating parameters of the equipment.

[0009] By adopting the above technical solution, the monitoring and acquisition unit of the visual monitoring module acquires the product transportation video stream in real time. After the data processing unit decomposes the image sequence using the inter-frame difference method, the template recognition unit accurately filters the product identification images based on the improved SIFT feature matching algorithm (introducing a dynamic sharpness threshold to adaptively adjust the matching parameters). The identification parsing unit uses the OCR-Tesseract engine to parse the identification code and aggregate the target location information. At the same time, the equipment monitoring unit integrates IMU and laser ranging data to monitor the status of the transportation equipment in real time. The tracking and tracing unit dynamically corrects the transportation parameters by combining the Kalman filter algorithm and outbound constraints. Through the collaboration of multiple algorithms, the high accuracy of identification recognition under complex lighting and occlusion scenarios is ensured, and the closed-loop feedback mechanism reduces the deviation of the transportation trajectory, significantly improving the reliability and traceability efficiency of the intelligent logistics system.

[0010] The present invention is further configured such that: the item transport module includes several sections of conveyor; the conveyor includes a single electric roller shaft and an equal number of electric rollers deployed on both sides of the electric roller shaft, for transporting products to the target location.

[0011] By adopting the above technical solution, a conveyor structure with a single electric roller shaft and symmetrically distributed electric rollers is used. The roller speed is dynamically adjusted by a PID control algorithm to achieve precise conveying. The product position is detected in real time by a vision camera and data acquisition is triggered. The transmission speed is adaptively adjusted by analyzing sensor feedback through an edge computing module. The product offset rate is reduced by a symmetrical power layout. The transmission efficiency is improved and energy consumption is reduced by real-time closed-loop control. At the same time, the mechanical structure is simplified and maintenance costs are reduced.

[0012] The present invention is further configured such that the tracking and tracing unit includes: The identification and labeling layer is used to identify all products in the product delivery image according to the target recognition algorithm and mark the product tracking boxes; The tracking and positioning layer is used to locate each product tracking frame according to the target location to obtain the real-time position of the product; An anomaly monitoring layer is used to determine anomalies in the product's operating status based on the product's form and real-time location, and to record the location of the anomaly. The instruction generation layer is used to generate parameter correction instructions based on the abnormal location and the equipment operating parameters of each conveyor segment; The operation correction layer is used to regulate the operating status of each conveyor in the goods transportation module according to the parameter correction instructions and the preset abnormal handling logic. The operation traceability layer is used to trace the operation process of a product based on the product identification code, thereby obtaining the product operation sequence. The outbound management layer is used to verify products based on preset outbound constraints. If the current product meets all outbound constraints, outbound permissions are granted to the current product.

[0013] The preset outbound constraints include complete product entry information, the product being located at the target location, and the product staying at the target location for a preset time limit.

[0014] By adopting the above technical solutions, YOLOv5 and other target recognition algorithms are used to identify and mark product transport images in real time, and DeepSORT multi-target tracking algorithm is used to achieve product positioning. Sliding window displacement threshold algorithm is used to detect abnormal positions, and parameter correction instructions are generated through decision tree. The transportation module is dynamically controlled based on PID control algorithm. The product identification code is traced throughout the entire operation by relying on hash traceability algorithm. Finally, the outbound constraints (information integrity, location matching and dwell time limit compliance) are verified by state machine to grant outbound permissions. The recognition and positioning accuracy is improved by multi-algorithm collaboration, the response time of abnormality detection is reduced, and the automation level of product transportation management is improved by combining real-time control and traceability mechanism, which significantly reduces the cost of manual intervention and ensures outbound compliance.

[0015] The present invention is further configured such that the regulation process of the operation correction layer includes: The parameter correction instruction is parsed according to the preset exception handling logic to obtain the instruction execution sequence and instruction correction parameters; The operating status of each conveyor section in the goods transportation module is monitored to obtain the rotation speed and rotation direction; Based on the distance between the real-time product location and the target location, the deviation compensation parameters are calculated by performing deviation compensation calculation on the instruction correction parameters. The rotation direction is adjusted according to the execution sequence of the instructions, and the rotation speed is corrected in conjunction with the deviation compensation parameters.

[0016] The operation process of the anomaly monitoring layer includes: The product tracking frame is used to monitor the delivery form of the product, and the product form similarity in adjacent product delivery images is calculated and compared with a preset form similarity threshold. If the product shape similarity is less than the shape similarity threshold, it is determined that the current product shape has changed, and the real-time position of the product in the adjacent product delivery image is determined. If the current real-time position of the product is not the target position, and the real-time position of the product remains unchanged, then the current operating state of the product is determined to be abnormal, and the abnormal position is recorded.

[0017] By adopting the above technical solution, YOLOv5 is used to detect product tracking frames in real time and calculate morphological similarity. When the similarity between adjacent frames is lower than a preset threshold (e.g., SSIM < 0.85) and the position deviates from the target, anomaly monitoring is triggered. A sliding window mechanism is used to analyze sudden changes in product displacement (e.g., Haversine distance > dynamic threshold) and record the abnormal position. The correction layer uses a PID control algorithm to parse parameter correction instructions and dynamically adjusts the rotation speed and direction of the transport module through a deviation compensation model. At the same time, Kalman filtering is used to predict and compensate for transmission delay. The outbound management layer uses a state machine to verify the integrity of product information, the matching degree of target position (Euclidean distance < 5cm), and the dwell time (≥ preset value). Finally, the entire process data is traced through a blockchain hash chain. By forming a closed loop of "detection-decision-control-verification", the anomaly response time is reduced, the positioning accuracy is improved, and the adaptive compensation algorithm improves the efficiency of transport correction. Combined with a decentralized traceability mechanism, the data is guaranteed to be tamper-proof, realizing fully automated and high-precision logistics management.

[0018] Secondly, the present invention also provides a product conveying method based on machine vision, which adopts the following technical solution: A machine vision-based product conveying method, applied to the product conveying system, includes: The system collects and decomposes the product's operational status, acquires several product delivery images, and marks the product's real-time location. The product delivery images are matched and filtered according to a preset identification template to obtain several product identification images; The target location is obtained by parsing and aggregating the same product identification code from the product identification images; Based on the product form, the real-time location of the product, and the target location, anomalies in the product's operating status are determined, and the abnormal locations are recorded. Monitor the operating status of the product operating equipment, obtain the equipment operating parameters, and generate parameter correction instructions based on the conveyor section corresponding to the abnormal position; According to the parameter correction instruction, the operating parameters of the product's operating equipment are adjusted according to the preset abnormality handling logic. The product's operation process is verified and traced based on preset outbound constraints and product identification codes.

[0019] By adopting the above technical solution, YOLOv5 is used to detect the product position in real time and the SSIM algorithm is used for template matching and filtering. The target position is determined by parsing the identifier code using CRNN. ​​The sliding window mechanism is used to calculate the morphological similarity between adjacent frames (an alarm is triggered when SSIM < 0.85) and Haversine distance displacement detection (dynamic threshold = historical average + 3σ) to realize anomaly judgment. The PID control algorithm generates parameter correction instructions to regulate the equipment speed, and Kalman filtering is used to compensate for transmission delay. Finally, the entire process data of the product is traced by the blockchain hash chain, and the outbound constraints (position deviation < 5cm, dwell time ≥ threshold) are verified by the state machine. By forming a vision-control-verification closed loop system, the abnormal response is reduced, the positioning accuracy is improved, and the transportation correction efficiency is improved. The decentralized traceability mechanism ensures that the data is tamper-proof and improves the automation rate of the whole process.

[0020] Thirdly, the present invention also provides an electronic device, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method described in the above scheme.

[0021] Fourthly, the present invention also provides a storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the product delivery method as described above.

[0022] In summary, the beneficial technical effects of the present invention are as follows: By using a visual recognition system to dynamically capture data during product operation in real time instead of photoelectric sensors, the problem of data loss can be improved, the standardized wiring harness of PLC can be eliminated, and the interaction costs of installation workers, electrical and mechanical personnel can be reduced significantly. When the product starts entering information at the input end, it can perform secondary information confirmation through vision to reduce packet loss and mistransmission issues. By sending instructions to the electric roller shaft through a vision solution to adjust the product's status, the need for human intervention is reduced, thus reducing the pressure on equipment operation and maintenance. Modular design and pre-configured connectors for all product segments eliminate the need for on-site electrical debugging, improving installation and deployment efficiency. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the structure of a product conveying system according to one embodiment of the present invention; Figure 2This is a schematic diagram of the structure of a visual monitoring module according to one embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a tracking and tracing unit according to one embodiment of the present invention; Figure 4 This is a schematic diagram of the product conveying control process according to one embodiment of the present invention; Figure 5 This is a schematic flowchart of a product conveying method according to one embodiment of the present invention. Detailed Implementation

[0024] The present invention will be further described in detail below with reference to the accompanying drawings.

[0025] Reference Figure 1 The present invention discloses a product conveying system based on machine vision, comprising: The visual monitoring module is used to identify and parse product identification codes, obtain the target location of the product, track and trace the operating status of the product, and control the operating status of each section of the conveyor in the goods transportation module. The item transportation module is used to transport the product according to the target location of the product; The remote control module is used to connect to and control the visual monitoring module and the goods transportation module based on the device access address, Wi-Fi account and Wi-Fi password.

[0026] Reference Figure 2 The visual monitoring module includes: The monitoring and acquisition unit is used to monitor and acquire the product's operating status to obtain the product's video stream. The data processing unit is used to perform frame decomposition on the product delivery video stream to obtain several product delivery images; The template recognition unit is used to match and filter the product delivery images one by one according to the preset identification template and the recognition clarity threshold to obtain a number of product identification images. The identifier parsing unit is used to parse the product identifier image and aggregate the same product identifier code to obtain the target location; The equipment monitoring unit is used to monitor the operating status of each section of the conveyor in the goods transportation module and obtain the equipment operating parameters; The tracking and tracing unit is used to track and locate the real-time position of several products and trace their operation according to preset outbound constraints, and to correct the operating parameters of the equipment.

[0027] Reference Figure 3 The tracking and tracing unit includes: The identification and labeling layer is used to identify all products in the product delivery image according to the target recognition algorithm and mark the product tracking boxes; The tracking and positioning layer is used to locate each product tracking frame according to the target location to obtain the real-time position of the product; An anomaly monitoring layer is used to determine anomalies in the product's operating status based on the product's form and real-time location, and to record the location of the anomaly. The instruction generation layer is used to generate parameter correction instructions based on the abnormal location and the equipment operating parameters of each conveyor segment; The operation correction layer is used to regulate the operating status of each conveyor in the goods transportation module according to the parameter correction instructions and the preset abnormal handling logic. The operation traceability layer is used to trace the operation process of a product based on the product identification code, thereby obtaining the product operation sequence. The outbound management layer is used to verify products based on preset outbound constraints. If the current product meets all outbound constraints, outbound permissions are granted to the current product.

[0028] The preset outbound constraints include complete product entry information, the product being located at the target location, and the product staying at the target location for a preset time limit.

[0029] The control process of the operation correction layer includes: The parameter correction instruction is parsed according to the preset exception handling logic to obtain the instruction execution sequence and instruction correction parameters; The operating status of each conveyor section in the goods transportation module is monitored to obtain the rotation speed and rotation direction; Based on the distance between the real-time product location and the target location, the deviation compensation parameters are calculated by performing deviation compensation calculation on the instruction correction parameters. The rotation direction is adjusted according to the execution sequence of the instructions, and the rotation speed is corrected in conjunction with the deviation compensation parameters.

[0030] The operation process of the anomaly monitoring layer includes: The product tracking frame is used to monitor the delivery form of the product, and the product form similarity in adjacent product delivery images is calculated and compared with a preset form similarity threshold. If the product shape similarity is less than the shape similarity threshold, it is determined that the current product shape has changed, and the real-time position of the product in the adjacent product delivery image is determined. If the current real-time position of the product is not the target position, and the real-time position of the product remains unchanged, then the current operating state of the product is determined to be abnormal, and the abnormal position is recorded.

[0031] The implementation principle of this embodiment is as follows: The monitoring and acquisition unit of the visual monitoring module acquires the product delivery video stream in real time. After the data processing unit decomposes the frames into image sequences, the template recognition unit matches and filters them against preset identifier templates based on the SSIM algorithm (similarity threshold > 0.85), extracts the product identifier image, and parses the identifier code using the CRNN algorithm to aggregate the target location. The tracking and tracing unit uses YOLOv5 to mark the product tracking box, combines it with the DeepSORT algorithm to achieve real-time multi-target positioning, and uses a sliding window mechanism to calculate the morphological similarity between adjacent frames (anomaly is triggered when SSIM < 0.8) and the Haversine distance to dynamically detect displacement abrupt changes (…). The threshold is calculated as historical average + 3σ. When a product's shape is abnormal and its position deviates from the target, the abnormal coordinates are recorded. The operation correction layer analyzes parameter correction instructions based on the PID control algorithm, predicts and compensates for transmission delays through Kalman filtering, and dynamically adjusts the speed and direction of the transportation module. At the same time, it combines real-time parameters fed back by the equipment monitoring unit to compensate for deviations. The outbound management layer verifies the integrity of product information, the accuracy of target position matching (Euclidean distance < 2cm), and the dwell time (≥ 10 seconds) based on the state machine. When the conditions are met, outbound permissions are granted, and the entire process data traceability is realized through the blockchain hash chain, forming an intelligent logistics system of "visual recognition - abnormal arbitration - dynamic control - closed-loop verification".

[0032] In this embodiment, the goods transportation module includes several conveyor sections; each conveyor includes a single electric roller shaft and an equal number of electric rollers deployed on both sides of the electric roller shaft, used to transport products to the target location. The remote control module is a WiFi remote control module; eliminating the need for traditional equipment to rely on photoelectric sensors to detect positioning signals for precise judgment. Example

[0033] Reference Figure 4 When the product is a material box: the material transportation module is equipped with only one electric roller shaft, eliminating the need for PLC standardized wiring harnesses, which reduces most of the installation, electrical and mechanical interaction costs; there are also no electrical and software interaction costs, thus improving efficiency; it can achieve a balanced distribution of the number of electric rollers in the entire transportation structure, and, due to the control scheme that does not use PLC wiring harnesses, the installation efficiency and effectiveness are greatly improved. The WiFi remote module is used as a control switch. It requires three pieces of configuration data: MAC address configuration (device ID), WiFi account and password, and connection within the local area network. Note that this connection is not in the cloud, as the cloud is too slow. A local area network can be set up within the factory where the device is located. The visual monitoring module uses cameras, and the number is determined by the camera's field of view. There is no fixed number of cameras; as long as the entire area of ​​the equipment can be covered, one or five cameras are sufficient. It depends entirely on the camera's field of view. Cameras can be installed by hanging, ceiling mounting, or direct reinforcement onto the equipment, without any restrictions.

[0034] Material is fed into the input end, and the barcode is scanned manually or identified by machine at the feeding port. The barcode on the material box is then identified. Once the barcode is successfully scanned, that is, after the material box reaches the transport structure, the system can know where the material box is going, such as to the first output end or the second output end. This is the content marked on the barcode on the material box. One material is fed into the input end at a time, but there can be multiple input ends.

[0035] A target tracking and positioning system is set up between the input and output ends to monitor the real-time status of the material box. This is achieved through a pre-installed vision camera. Furthermore, the vision camera can also monitor the rotation of the electric roller shaft. Specifically, the electric roller shaft initially operates at a constant speed. If a problem occurs at some point, such as the material box becoming misaligned or jammed due to excessive weight, the material box will never reach its designated location. At this time, the vision camera will monitor the operating status of the material box, as well as the rotation speed and number of revolutions of the electric roller shaft. If it is still stuck, the vision camera will send a command to the system to make the electric roller shaft reverse several times and then rotate normally to see if it can be improved. If the material box is still stuck, an alarm will be triggered, and then manual handling will be required. In addition, the target tracking and positioning module can also control the electric roller shaft to rotate at a differential speed, so as to avoid multiple material boxes getting too close together and forming overlapping images, which would prevent the material boxes from being visually identified. Furthermore, for factories that need to install monitoring systems during operation, the target tracking and positioning module of a vision camera can replace the monitoring system. In addition, lost bins can be traced through the operation process to avoid loss of packages; Finally, it also ensures the timely release of material bins. Conditions are set so that after a material bin reaches its target location, it can only be released if the following three conditions are met, preventing the bin from being released before the information is actually entered into the database. Specifically, the three conditions are: 1. The local area network has identified and completed the information entry, at which point the information is complete; 2. The visual system can recognize that the material bin is at this location; 3. A set dwell time for the material bin at the target location is set, and condition three is only considered complete after the dwell time is reached. Therefore, the material bin is only allowed to leave after all three conditions are met, reducing the problem of the material bin leaving without data being entered into the database.

[0036] Reference Figure 5 A machine vision-based product conveying method, applied to the machine vision-based product conveying system, includes: S1: Collect and frame decompose the product's operating status, acquire several product delivery images, and mark the product's real-time location. S2: Match and filter the product delivery images according to the preset identification template to obtain several product identification images; S3: Parse and aggregate the same product identification code in the product identification image to obtain the target location; S4: Based on the product form, the real-time location of the product, and the target location, make an anomaly judgment on the product's operating status and record the anomaly location. S5: Monitor the operating status of the product operating equipment, obtain the equipment operating parameters, and generate parameter correction instructions based on the conveyor section corresponding to the abnormal position; S6: Adjust the equipment operating parameters of the product operating equipment according to the preset abnormality handling logic based on the parameter correction instruction; S7: Verify and trace the product's operation process based on preset outbound constraints and product identification codes.

[0037] The implementation principle of this embodiment is as follows: The visual monitoring module collects product operation video and decomposes it into image sequences. The real-time position is marked by the YOLOv5 algorithm. After filtering and marking the images using SSIM template matching, the target position is generated by parsing the identification code through CRNN. ​​The product shape similarity is calculated based on the sliding window (e.g., shape change is triggered when SSIM < 0.8) and the displacement mutation is dynamically detected by Haversine distance (threshold = historical average + 3σ). When the shape is abnormal and deviates from the target position, the abnormal coordinates are recorded. The equipment monitoring unit synchronously obtains the operating parameters (speed and direction) of the transportation module, generates correction instructions through the PID control algorithm and combines Kalman filtering to compensate for transmission delay and adjust the equipment parameters. The outbound management layer uses a state machine to verify the integrity of product information, the accuracy of target position matching (Euclidean distance < 2cm) and the dwell time limit (≥ 10 seconds). If they meet the requirements, outbound permission is granted, and the whole process is traced through the blockchain hash chain.

[0038] An electronic device, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement any of the methods described in the above scheme.

[0039] A storage medium storing at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the product data management method as described above.

[0040] The embodiments described herein are preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A product conveying system based on machine vision, comprising: The visual monitoring module is used to identify and parse product identification codes, obtain the target location of the product, track and trace the operating status of the product, and control the operating status of each section of the conveyor in the goods transportation module. The item transportation module is used to transport the product according to the target location of the product; The remote control module is used to connect to and control the visual monitoring module and the goods transportation module based on the device access address, Wi-Fi account and Wi-Fi password.

2. The product conveying system based on machine vision according to claim 1, characterized in that, The visual monitoring module includes: The monitoring and acquisition unit is used to monitor and acquire the product's operating status to obtain the product's video stream. The data processing unit is used to perform frame decomposition on the product delivery video stream to obtain several product delivery images; The template recognition unit is used to match and filter the product delivery images one by one according to the preset identification template and the recognition clarity threshold to obtain a number of product identification images. The identifier parsing unit is used to parse the product identifier image and aggregate the same product identifier code to obtain the target location; The equipment monitoring unit is used to monitor the operating status of each section of the conveyor in the goods transportation module and obtain the equipment operating parameters; The tracking and tracing unit is used to track and locate the real-time position of several products and trace their operation according to preset outbound constraints, and to correct the operating parameters of the equipment.

3. The product conveying system based on machine vision according to claim 1, characterized in that, The goods transport module includes several sections of conveyor; each conveyor includes a single electric roller shaft and an equal number of electric rollers deployed on both sides of the electric roller shaft, for transporting products to the target location.

4. The product conveying system based on machine vision according to claim 2, characterized in that: The tracking and tracing unit includes: The identification and labeling layer is used to identify all products in the product delivery image according to the target recognition algorithm and mark the product tracking boxes; The tracking and positioning layer is used to locate each product tracking frame according to the target location to obtain the real-time position of the product; An anomaly monitoring layer is used to determine anomalies in the product's operating status based on the product's form and real-time location, and to record the location of the anomaly. The instruction generation layer is used to generate parameter correction instructions based on the abnormal location and the equipment operating parameters of each conveyor segment; The operation correction layer is used to regulate the operating status of each conveyor in the goods transportation module according to the parameter correction instructions and the preset abnormal handling logic. The operation traceability layer is used to trace the operation process of a product based on the product identification code, thereby obtaining the product operation sequence. The outbound management layer is used to verify products based on preset outbound constraints. If the current product meets all outbound constraints, outbound permissions are granted to the current product.

5. The product conveying system based on machine vision according to claim 2 or 4, characterized in that: The preset outbound constraints include complete product entry information, the product being located at the target location, and the product staying at the target location for a preset time limit.

6. The product conveying system based on machine vision according to claim 4, characterized in that: The control process of the operation correction layer includes: The parameter correction instruction is parsed according to the preset exception handling logic to obtain the instruction execution sequence and instruction correction parameters; The operating status of each conveyor section in the goods transportation module is monitored to obtain the rotation speed and rotation direction; Based on the distance between the real-time product location and the target location, the deviation compensation parameters are calculated by performing deviation compensation calculation on the instruction correction parameters. The rotation direction is adjusted according to the execution sequence of the instructions, and the rotation speed is corrected in conjunction with the deviation compensation parameters.

7. The product conveying system based on machine vision according to claim 4, characterized in that: The operation process of the anomaly monitoring layer includes: The product tracking frame is used to monitor the delivery form of the product, and the product form similarity in adjacent product delivery images is calculated and compared with a preset form similarity threshold. If the product shape similarity is less than the shape similarity threshold, it is determined that the current product shape has changed, and the real-time position of the product in the adjacent product delivery image is determined. If the current real-time position of the product is not the target position, and the real-time position of the product remains unchanged, then the current operating state of the product is determined to be abnormal, and the abnormal position is recorded.

8. A product conveying method based on machine vision, characterized in that, include: The system collects and decomposes the product's operational status, acquires several product delivery images, and marks the product's real-time location. The product delivery images are matched and filtered according to a preset identification template to obtain several product identification images; The target location is obtained by parsing and aggregating the same product identification code from the product identification images; Based on the product form, the real-time location of the product, and the target location, anomalies in the product's operating status are determined, and the abnormal locations are recorded. Monitor the operating status of the product operating equipment, obtain the equipment operating parameters, and generate parameter correction instructions based on the conveyor section corresponding to the abnormal position; According to the parameter correction instruction, the operating parameters of the product's operating equipment are adjusted according to the preset abnormality handling logic. The product's operation process is verified and traced based on preset outbound constraints and product identification codes.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the method as described in claim 8.

10. A storage medium storing at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the product delivery method of claim 8.

Images