Logistics transfer site early warning management method, device, computer equipment and storage medium

Abstract

The present application relates to a logistics transfer site early warning management method, device, computer equipment and storage medium. The method includes: obtaining the current video frame; performing target detection on the current video frame to obtain the sorting status of each cargo in the current video frame; performing statistics on the cargo with a normal sorting status in the current video frame to obtain the total quantity of cargo corresponding to the current video frame; determining the quantity information of the current time period based on the total quantity of cargo corresponding to each video frame in the current time period, and predicting the quantity information of the next time period based on the quantity information of the current time period; when it is determined that the incoming cargo in the next time period is abnormal based on the quantity information of the current time period and the quantity information of the next time period, triggering an early warning message. The use of this method can improve the efficiency of cargo sorting.

Description

Technical Field

[0001] The present application relates to the technical field of logistics transfer site management, and in particular to a logistics transfer site early warning management method, device, computer equipment and storage medium. Background Art

[0002] With the rapid development of e-commerce and the logistics industry in recent years, centralized cargo sorting is increasingly being used. To improve sorting and handling efficiency, the logistics industry has steadily increased its investment in automated equipment. However, managing logistics transfer sites and optimizing the management of equipment, cargo, and personnel within these sites has long been a challenge for the industry. Currently, management of logistics transfer sites typically involves post-event accountability after an abnormality occurs, but inefficient cargo sorting is often not promptly identified, resulting in reduced efficiency. Summary of the Invention

[0003] Based on this, it is necessary to provide a logistics transfer site early warning management method, device, computer equipment and storage medium that can improve the efficiency of cargo sorting in response to the above technical problems.

[0004] A logistics transfer site early warning management method, the method comprising:

[0005] Get the current video frame;

[0006] Performing target detection on the current video frame to obtain a sorting status of each item in the current video frame;

[0007] Counting the goods in the normal sorting state in the current video frame to obtain a total quantity corresponding to the current video frame;

[0008] Determine the piece quantity information of the current time period according to the total piece quantity corresponding to each video frame in the current time period, and predict the piece quantity information of the next time period according to the piece quantity information of the current time period;

[0009] When it is determined based on the package quantity information of the current period and the package quantity information of the next period that the incoming packages of the next period are frequent, an early warning message is triggered.

[0010] In one embodiment, performing target detection on the current video frame to obtain the sorting status of each item in the current video frame includes:

[0011] Performing target detection on the current video frame to obtain a sorting status and a cargo detection frame of each cargo in the current video frame;

[0012] The method further comprises:

[0013] When the current video frame includes goods with an abnormal sorting status, an alarm message is triggered; the alarm message includes the current video frame and a goods detection frame corresponding to the goods with an abnormal sorting status.

[0014] In one embodiment, performing target detection on the current video frame to obtain the sorting status and the cargo detection frame of each cargo in the current video frame includes:

[0015] The current video frame is input into a trained target detection model for target detection to obtain the sorting status and the target detection frame of each item in the current video frame.

[0016] In one embodiment, predicting the quantity information of the next time period based on the quantity information of the current time period includes:

[0017] The quantity information of the current time period is input into the trained quantity prediction model for prediction to obtain the quantity information of the next time period.

[0018] In one embodiment, the step of training the quantity prediction model includes:

[0019] Get the total number of pieces corresponding to each video frame in multiple historical periods;

[0020] Determine the quantity information of the corresponding historical period according to the total quantity corresponding to each video frame in each historical period;

[0021] For each pair of adjacent historical periods, the quantity information of the previous historical period is used as the input feature, and the quantity information of the next historical period is used as the expected output feature for model training to obtain a trained quantity prediction model.

[0022] In one embodiment, the method further comprises:

[0023] When it is determined based on the package volume information of the current period and the package volume information of the next period that the distribution of incoming packages in the next period is abnormal, the number of personnel and equipment required in the next period are dynamically determined based on the package volume information of the next period;

[0024] Dynamically deploying staff for the next time period based on the number of staff in the current time period and the number of staff required for the next time period;

[0025] Dynamically allocate working equipment in the next time period according to the number of equipment in the current time period and the number of equipment required in the next time period.

[0026] In one embodiment, there are multiple current video frames; the multiple video frames are synchronously acquired from corresponding video streams; the multiple video streams are synchronously acquired by corresponding video acquisition devices; and execution is started from the step of performing target detection on the current video frame to obtain the sorting status of each cargo in the current video frame based on the multiple current video frames in parallel.

[0027] A logistics transfer site early warning management device, the device comprising:

[0028] Acquisition module, used to obtain the current video frame;

[0029] a detection module, configured to perform target detection on the current video frame to obtain a sorting status of each item in the current video frame;

[0030] A statistics module, configured to count the goods in the normal sorting state in the current video frame to obtain a total quantity of goods corresponding to the current video frame;

[0031] A prediction module, configured to determine the quantity information of a current period based on the total quantity of video frames corresponding to the current period, and predict the quantity information of a next period based on the quantity information of the current period;

[0032] The early warning module is used to trigger an early warning message when it is determined that the distribution of incoming items in the next time period is abnormal based on the item quantity information of the current time period and the item quantity information of the next time period.

[0033] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0034] Get the current video frame;

[0035] Performing target detection on the current video frame to obtain a sorting status of each item in the current video frame;

[0036] Counting the goods in the normal sorting state in the current video frame to obtain a total quantity corresponding to the current video frame;

[0037] Determine the piece quantity information of the current time period according to the total piece quantity corresponding to each video frame in the current time period, and predict the piece quantity information of the next time period according to the piece quantity information of the current time period;

[0038] When it is determined that the distribution of incoming parcels in the next time period is abnormal based on the parcel quantity information of the current time period and the parcel quantity information of the next time period, an early warning message is triggered.

[0039] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0040] Get the current video frame;

[0041] Performing target detection on the current video frame to obtain a sorting status of each item in the current video frame;

[0042] Counting the goods in the normal sorting state in the current video frame to obtain a total quantity corresponding to the current video frame;

[0043] Determine the piece quantity information of the current time period according to the total piece quantity corresponding to each video frame in the current time period, and predict the piece quantity information of the next time period according to the piece quantity information of the current time period;

[0044] When it is determined that the distribution of incoming parcels in the next time period is abnormal based on the parcel quantity information of the current time period and the parcel quantity information of the next time period, an early warning message is triggered.

[0045] The above-mentioned logistics transfer site early warning management method, device, computer equipment and storage medium detect the sorting status of each cargo in the current video frame in real time to determine the total number of pieces with normal sorting status in the current video frame according to the sorting status of each cargo, determine the piece quantity information of the current time period based on the total piece quantity corresponding to each video frame in the current time period in which the current video frame is located, predict the piece quantity information of the next time period based on the piece quantity information of the current time period, and determine the incoming piece situation in the next time period based on the piece quantity information of the current time period and the piece quantity information of the next time period, so as to trigger early warning information when the incoming piece in the next time period is abnormal, so as to instruct the management personnel to timely deploy staff and work equipment through the early warning information, thereby improving the efficiency of cargo sorting. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a flow chart of a logistics transfer site early warning management method in one embodiment;

[0047] Figure 2 A schematic flow chart of a logistics transfer site early warning management method in another embodiment;

[0048] Figure 3 This is a schematic diagram of the structure of a logistics transfer site early warning management system in one embodiment;

[0049] Figure 4 A schematic diagram of the principle of a logistics transfer site early warning management method in one embodiment;

[0050] Figure 5 This is a structural block diagram of a logistics transfer site early warning management device in one embodiment;

[0051] Figure 6 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0053] In one embodiment, Figure 1 As shown, a logistics transfer site early warning management method is provided. This embodiment uses the method applied to a server as an example for illustration. It is understandable that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0054] Step 102: Get the current video frame.

[0055] Specifically, the server extracts a video frame from the video stream as the current video frame. It can be understood that the server regularly extracts video frames from the video stream according to a preset period, uses the video frame extracted each time as the current video frame, and after each extraction of the current video frame, based on the extracted current video frame, executes the corresponding process in accordance with the logistics transfer site early warning management method provided in one or more embodiments of the present application. The preset period can be customized according to actual conditions, such as a preset period of extracting a video frame as the current video frame every two video frames, or a preset period of extracting each video frame from the video stream in sequence as the current video frame, that is, extracting the current video frame from the video stream in real time.

[0056] In one embodiment, the video stream is captured by a video capture device, such as a camera deployed in a logistics transfer station, and sent to the server. The camera is such as an industrial camera.

[0057] In one embodiment, a logistics transfer site is deployed with multiple video capture devices, which synchronously capture video streams. The server receives the video streams synchronously captured and sent by the multiple video capture devices, extracts video frames in parallel from the multiple synchronously received video streams as current video frames, and executes the relevant processes in the logistics transfer site early warning management method provided by this application based on each current video frame extracted in parallel. In this way, based on the current video frame corresponding to each video capture device, the distribution of incoming parcels in the next time period within the logistics transfer site area covered by the field of view of the corresponding video capture device is determined in parallel. When the distribution of incoming parcels in the next time period within a certain logistics transfer site area is abnormal, a corresponding early warning information is triggered for the logistics transfer site area, so that the management personnel can promptly deploy the staff and work equipment in the logistics transfer site area based on the early warning information, thereby improving the management efficiency of personnel and equipment in the logistics transfer site, thereby improving the efficiency of cargo sorting.

[0058] Step 104: Perform target detection on the current video frame to obtain the sorting status of each item in the current video frame.

[0059] The sorting state refers to the sorting state of the goods contained in the video frame when the video frame is collected, that is, it refers to the state of the goods at the moment the corresponding video frame is collected during the goods sorting process. The sorting state includes normal items and abnormal items. Abnormal items may include stuck items and dropped items (such as goods being squeezed off the goods sorting belt). When collecting a video frame, if one or more goods in the video frame are in a normal sorting state, the sorting state of the one or more goods is determined to be a normal item. If one or more goods in the video frame are in an abnormal sorting state such as stuck items or dropped items, the sorting state of the one or more goods is determined to be an abnormal item.

[0060] Specifically, the server performs target detection on the acquired current video frame to extract each cargo from the current video frame and determine the sorting status of each cargo in the current video frame.

[0061] In one embodiment, the server performs object detection on the current video frame using a trained object detection model to obtain the sorting status of each item in the current video frame.

[0062] In one embodiment, the server performs object detection on the current video frame and also obtains a corresponding cargo detection frame for each cargo item in the current video frame, so that the location of the corresponding cargo item in the current video frame can be determined based on the cargo detection frame. It will be appreciated that when the server performs object detection on the current video frame using a trained object detection model, it also obtains a corresponding cargo detection frame for each cargo item in the current video frame.

[0063] Step 106: Count the goods in the normal sorting state in the current video frame to obtain the total quantity corresponding to the current video frame.

[0064] The total number of pieces corresponding to the current video frame refers to the total number of pieces of goods with a normal sorting status in the current video frame, which can also be understood as the total number of pieces of goods or the statistical number of pieces.

[0065] Specifically, the server filters out goods with a normal sorting status from the current video frame according to the sorting status of each good in the current video frame, counts the total number of the filtered goods, and determines the total number of goods obtained by counting as the total number of goods corresponding to the current video frame.

[0066] Step 108 : determining the piece quantity information of the current time period according to the total piece quantity corresponding to each video frame in the current time period, and predicting the piece quantity information of the next time period according to the piece quantity information of the current time period.

[0067] Among them, the piece quantity information is information used to characterize the piece quantity situation in a single time period, which can be the total piece quantity or the average piece quantity. For example, the piece quantity information of the current time period can refer to the total piece quantity in the current time period, or it can refer to the average piece quantity in the current time period.

[0068] The current time period refers to the time period in which the current video frame is located. The current time period can specifically refer to a time interval that ends with the current video frame and includes a preset number of video frames, or it can refer to a time interval that ends with the current video frame and has a preset duration. The next time period corresponds to the current time period, and each current time period corresponds to a next time period. The next time period refers to the time period next to the current time period. The next time period can specifically refer to a time interval that starts with the next video frame after the current video frame and includes a preset number of video frames, or it can refer to a time interval that starts with the next video frame after the current video frame and has a preset duration. The number of video frames included in the current time period and the next time period is the same, and accordingly, the time length covered by the current time period and the next time period is the same. The preset number can be customized, such as 100 frames, and the preset duration can be customized, such as 3 minutes.

[0069] It can be understood that the next video frame of the current video frame refers to a video frame in the video stream that has a later acquisition time than the current video frame and will be selected as the current video frame in the next cycle. For example, assuming that the current video frame is the 100th frame in the video stream, if the preset cycle is to extract one video frame as the current video frame every two video frames, then the next video frame of the current video frame is the 103rd frame. If the preset cycle is to extract each video frame from the video stream in sequence as the current video frame, then the next video frame of the current video frame is the 101st frame.

[0070] Specifically, the server determines the current time period in which the current video frame is located, and the various video frames included in the current time period based on the current video frame, obtains the total number of pieces corresponding to each video frame in the current time period, determines the piece quantity information corresponding to the current time period based on the total number of pieces corresponding to each video frame in the current time period, and predicts the piece quantity information for the next time period based on the piece quantity information of the current time period. It can be understood that the total number of pieces corresponding to each video frame in the current time period is the total number of pieces statistically obtained and stored corresponding to the current video frame during the process of extracting the current video frame from the video stream according to a preset period and executing the corresponding process based on the extracted current video frame in accordance with the logistics transfer site early warning management method provided in this application.

[0071] In one embodiment, the server sums up the total number of pieces corresponding to each video frame in the current time period to obtain the total number of pieces in the current time period, or averages the total number of pieces corresponding to each video frame in the current time period to obtain the average number of pieces in the current time period, and determines the piece quantity information of the current time based on the total number of pieces or the average number of pieces in the current time period.

[0072] In one embodiment, the server uses a trained volume prediction model to predict the volume for the next period based on the volume for the current period. It is understood that if the volume for the current period is the total volume, the predicted volume for the next period will also be the total volume.

[0073] Step 110 : When it is determined that the incoming parcels in the next period are abnormal based on the parcel quantity information of the current period and the parcel quantity information of the next period, an early warning message is triggered.

[0074] Specifically, after the server predicts the quantity information of the next time period, it determines the incoming parcels of the next time period based on the quantity information of the current time period and the quantity information of the next time period, that is, it determines whether the incoming parcels of the next time period are abnormal. When it is determined that the incoming parcels of the next time period are abnormal, the server triggers the generation of an early warning message and sends the generated early warning message to the terminal corresponding to the management personnel, so as to instruct the management personnel to dynamically deploy staff and work equipment through the terminal, so that the dynamically deployed staff and work equipment can adapt to the abnormal incoming parcel volume in the next time period, thereby improving the efficiency of cargo sorting. The abnormal incoming parcels can specifically be an abnormal quantity of incoming parcels or an abnormal distribution of incoming parcels.

[0075] In one embodiment, when the quantity information is the average quantity, if the average quantity of the next time period is compared with the average quantity of the current time period, and the fluctuating quantity (increasing or decreasing quantity) is too large, the incoming items of the next time period are judged to be abnormal. Specifically, the server calculates the fluctuating quantity based on the average quantity of the current time period and the average quantity of the next time period. If the quotient between the fluctuating quantity and the average quantity of the current time period is greater than or equal to the fluctuation threshold, the incoming items of the next time period are judged to be abnormal. The fluctuation threshold can be customized, such as 50%. Similarly, when the quantity information is the total quantity, if the total quantity of the next time period is compared with the total quantity of the current time period, and the fluctuating quantity (increasing or decreasing quantity) is too large, the incoming items of the next time period are judged to be abnormal.

[0076] In one embodiment, when it is determined that the incoming messages in the next time period are abnormal, the server triggers an early warning instruction and sends the early warning instruction to the early warning device configured for the corresponding video acquisition device to instruct the early warning device to issue an early warning.

[0077] In one embodiment, if multiple video capture devices are deployed in the logistics transfer yard, when it is determined based on the current video frame corresponding to a single video capture device and in accordance with the relevant process provided in this application that the logistics transfer yard area covered by the video capture device has abnormal incoming parcels in the next time period, the early warning information triggered and generated by the server will also carry the device identification of the video capture device, so that the logistics transfer yard area with abnormal incoming parcels in the next time period can be determined based on the device identification in the early warning information, and the video stream captured by the corresponding video capture device can be retrieved based on the device identification, so that the staff and work equipment can be deployed in a targeted manner.

[0078] The above-mentioned logistics transfer site early warning management method detects the sorting status of each cargo in the current video frame in real time to determine the total number of pieces with normal sorting status in the current video frame according to the sorting status of each cargo, determines the piece quantity information of the current time period based on the total piece quantity corresponding to each video frame in the current time period in which the current video frame is located, predicts the piece quantity information of the next time period based on the piece quantity information of the current time period, and determines the incoming piece situation in the next time period based on the piece quantity information of the current time period and the piece quantity information of the next time period, so as to trigger early warning information when the incoming piece in the next time period is abnormal, so as to instruct the management personnel to timely deploy staff and work equipment through the early warning information, thereby improving the efficiency of cargo sorting.

[0079] In one embodiment, step 104 includes: performing target detection on the current video frame to obtain the sorting status and cargo detection frame of each cargo in the current video frame; the logistics transfer station early warning management method also includes: triggering an alarm message when the current video frame includes cargo with an abnormal sorting status; the alarm message includes the current video frame and the cargo detection frame corresponding to the cargo with an abnormal sorting status.

[0080] The cargo detection frame is used to identify or determine the location of the cargo within the current video frame. Specifically, it can be the minimum rectangle that can cover a single cargo item within the current video frame. The cargo detection frame can be characterized by cargo detection frame parameters, which uniquely identify the cargo detection frame. These parameters include, for example, the coordinates of the cargo detection frame's center point, the distance between the center point and at least one vertex or two adjacent sides of the cargo detection frame, and the coordinates of any two diagonal vertices of the cargo detection frame.

[0081] Specifically, the server performs target detection on the current video frame to determine the sorting status and cargo detection frame corresponding to each cargo in the current video frame. Furthermore, the server detects whether there are cargoes with abnormal sorting status in the current video frame based on the sorting status corresponding to each cargo in the current video frame. When it is determined that there are cargoes with abnormal sorting status in the current video frame, an alarm message is generated based on the cargo detection frame corresponding to the cargoes with abnormal sorting status and the current video frame. The server sends the triggered alarm message to the terminal corresponding to the management personnel, so as to instruct the management personnel to handle the cargo sorting abnormality in a timely manner through the terminal, so as to avoid the problem of untimely discovery or untimely handling of cargo sorting abnormalities.

[0082] In one embodiment, the alarm information may also include the acquisition timestamp corresponding to the current video frame, the video stream identifier of the video stream in which the current video frame is located, or the device identifier of the video acquisition device using the current video frame, and the actual location information of the goods with abnormal sorting status in the logistics transfer yard.

[0083] In one embodiment, when the current video frame is determined to include goods in an abnormal sorting status, the server triggers the generation of an alarm instruction and transmits the alarm instruction to the alarm device corresponding to the video capture device used to capture the current video frame, instructing the alarm device to issue an alarm. It is understood that when multiple video capture devices are deployed in a logistics transfer station, the early warning device and alarm device configured for each video capture device can be the same or different devices, without specific limitations here.

[0084] In the above embodiment, when there are goods with abnormal sorting status in the current video frame, the goods detection frame corresponding to the goods with abnormal sorting status and the alarm information of the current video frame are triggered, so that based on the alarm information, the goods in the abnormal sorting status can be quickly and accurately determined, and the position of the goods in the current video frame can be obtained, so that the goods in the abnormal sorting status can be processed accurately and timely, thereby improving the goods sorting efficiency.

[0085] In one embodiment, target detection is performed on the current video frame to obtain the sorting status and the detection frame of each cargo in the current video frame, including: inputting the current video frame into a trained target detection model to perform target detection, and obtaining the sorting status and the detection frame of each cargo in the current video frame.

[0086] The object detection model is trained based on a pre-acquired first training sample set and is capable of detecting the sorting status and detection frame of each item in a video frame. The first training sample set includes sample video frames, and the sorting status and detection frame of each item in the sample video frames.

[0087] Specifically, the server obtains multiple sample video frames, annotates each sample video frame to obtain the sorting status and detection frame of each item in each sample video frame, and then generates a first training sample set based on the multiple sample video frames and the sorting status and detection frame of each item in each sample video frame. Model training is then performed based on the first training sample set to obtain a trained object detection model. During model training, the server uses the sample video frames in the first training sample set as input features and the sorting status and detection frame of each item in the sample video frame as the desired output features for model training.

[0088] In one embodiment, the machine learning algorithm involved in training the target detection model includes but is not limited to YOLOv5 (a target detection network). The target detection model obtained by training based on YOLOv5 can be understood as a deep target detection model.

[0089] In one embodiment, the server obtains one or more historical video streams and extracts multiple sample video frames from the one or more historical video streams. It can be understood that if corresponding video acquisition devices are deployed in the logistics transfer yard for each logistics sorting link, the same or different target detection models can be used to perform target detection on the current video frames corresponding to each of the multiple video acquisition devices. If the same target detection model is used to perform target detection on the current video frames corresponding to each video acquisition device, then during the model training process, the server needs to obtain the historical video streams collected by each video acquisition device so that the first training sample set includes sample video frames under each logistics sorting link. In this way, the target detection model trained based on the first training sample set can be applied to target detection under each logistics sorting link. If different target detection models are used to perform target detection on the current video frames corresponding to each video acquisition device, then during the model training process, the sample video frames in the first training sample set used to train the target detection model are extracted from the historical video streams collected by the corresponding video acquisition devices.

[0090] In the above embodiment, by performing target detection on the current video frame using the trained target detection model, the sorting status and the target detection frame of each item in the current video frame can be determined quickly and accurately.

[0091] In one embodiment, predicting the quantity information of the next time period based on the quantity information of the current time period includes: inputting the quantity information of the current time period into a trained quantity prediction model for prediction to obtain the quantity information of the next time period.

[0092] The volume prediction model is trained based on a pre-acquired second training sample set and can be used to predict the volume information for the next period based on the volume information for the current period. The second training sample set includes multiple historical periods and the volume information for each historical period. For each historical period in the second training sample set, there is at least one adjacent historical period in the second training sample set, and two adjacent historical periods can form a pair of adjacent historical periods.

[0093] It can be understood that each pair of adjacent historical periods must at least meet the following conditions: the video frames in the two historical periods are derived from the same historical video stream, and the first video frame in the latter historical period is the next video frame of the last video frame in the previous historical period. The previous historical period corresponds to the latter historical period, and the latter historical period in a pair of adjacent historical periods can also be used as the previous historical period in another pair of adjacent historical periods. Taking the preset period of extracting each video frame from the video stream as the current video frame as an example, assuming that the video frames covered by the previous historical period are frames 1 to 100, then the video frames covered by the latter historical period are frames 101 to 200, and so on. If the video frames covered by the previous historical period are frames 101 to 200, then the video frames covered by the latter historical period are frames 201 to 300.

[0094] Specifically, after the server determines the quantity information of the current time period based on the total quantity corresponding to each video frame in the current time period, it inputs the quantity information of the current time period into the trained quantity prediction model, and uses the quantity prediction model to predict the quantity information of the next time period based on the quantity information of the current time period.

[0095] In the above embodiment, the trained package quantity prediction model can be used to quickly and accurately predict the package quantity information for the next time period based on the package quantity information for the current time period.

[0096] In one embodiment, the training steps of the piece quantity prediction model include: obtaining the total piece quantity corresponding to each video frame in multiple historical time periods; determining the piece quantity information of the corresponding historical time period based on the total piece quantity corresponding to each video frame in each historical time period; for each pair of adjacent historical time periods, using the piece quantity information of the previous historical time period as the input feature and the piece quantity information of the next historical time period as the expected output feature to train the model, thereby obtaining a trained piece quantity prediction model.

[0097] Specifically, the server obtains one or more historical video streams and extracts video frames from multiple historical time periods. Each of the multiple historical time periods has at least one adjacent historical time period within the extracted multiple historical time periods. For each of the extracted video frames from the multiple historical time periods, the server calculates the total number of pieces of goods with a normal sorting status in each video frame and uses this as the total number of pieces corresponding to the video frame. This allows the server to obtain the total number of pieces corresponding to each video frame within the multiple historical time periods. For each historical time period, the server obtains piece quantity information for that historical time period based on the total number of pieces corresponding to each video frame within that historical time period. Based on the piece quantity information for the extracted multiple historical time periods, the server obtains a second training sample set. Thus, the second training sample set includes multiple pairs of adjacent historical time periods. Furthermore, for each pair of adjacent historical time periods in the second training sample set, the server performs model training using the piece quantity information of the preceding historical time period as input features and the piece quantity information of the succeeding historical time period as desired output features, thereby obtaining a trained piece quantity prediction model.

[0098] In one embodiment, the quantity prediction model is obtained based on the second training sample set and the initial model hyperparameter training. It can be understood that, similar to the method of obtaining the second training sample set, the server obtains a test sample set, and after the training based on the second training sample set is completed, the quantity prediction model obtained by training based on the second training sample set is tested with the test sample set to obtain the quantity information prediction value of each subsequent historical period in the test sample set, and the predicted value of the quantity information of each subsequent historical period is compared with the actual quantity information corresponding to the subsequent historical period in the test sample set, and the mean square error between the two is calculated, and the hyperparameters of the trained quantity prediction model are adjusted based on the mean square error, and the quantity prediction model after the hyperparameter adjustment is determined as the trained quantity prediction model.

[0099] In one embodiment, the machine learning algorithms involved in training the volume prediction model include, but are not limited to, XGBoost (eXtreme Gradient Boosting).

[0100] In the above embodiment, the piece quantity prediction model is pre-trained so that in the logistics transit site early warning management process, the piece quantity information of the next time period can be quickly and accurately predicted based on the pre-trained piece quantity prediction model, and the prediction efficiency of the piece quantity information can be improved. When judging whether to trigger the early warning information based on the predicted piece quantity information, the early warning efficiency can be improved, thereby improving the cargo sorting efficiency.

[0101] In one embodiment, the logistics transit site early warning management method also includes: when it is determined that the incoming parcels in the next period are abnormal based on the parcel quantity information of the current period and the parcel quantity information of the next period, the number of personnel and the number of equipment required in the next period are dynamically determined based on the parcel quantity information of the next period; the staff in the next period are dynamically deployed based on the number of personnel in the current period and the number of personnel required in the next period; the working equipment in the next period is dynamically deployed based on the number of equipment in the current period and the number of equipment required in the next period.

[0102] The number of people in the current time period is calculated by averaging the number of people in each video frame within the current time period. Correspondingly, the number of devices in the current time period is calculated by averaging the number of devices in each video frame within the current time period. The number of people and devices in each video frame within the current time period can be obtained by detecting the corresponding video frames or by dynamically querying management data (such as personnel management data and equipment management data) of the corresponding logistics sorting process.

[0103] For example, assuming that the number of personnel and the number of equipment in the current period are 2 and 3 respectively, if the number of personnel and the number of equipment required in the next period are 1 and 2 respectively, then the number of personnel in the next period will be reduced to 1, and the number of working equipment in the next period will be reduced to 2. If the number of personnel and the number of equipment required in the next period are 5 and 8 respectively, then the number of personnel in the next period will be increased to 5, and the number of working equipment in the next period will be increased to 8.

[0104] In one embodiment, when the number of incoming packages in the next period is determined to be abnormal based on the package volume information of the current period and the package volume information of the next period, the server dynamically determines the number of personnel required for the next period based on the corresponding relationship between the package volume and the number of personnel, and dynamically determines the number of equipment required for the next period based on the corresponding relationship between the package volume and the number of equipment. The corresponding relationship between the package volume and the number of personnel, and the corresponding relationship between the package volume and the number of equipment, are obtained based on the analysis of a large amount of historical data.

[0105] In one embodiment, when the incoming parcels for the next period are determined to be abnormal, the server will also obtain the average efficiency of the personnel in the corresponding logistics sorting process and dynamically determine the number of personnel required for the next period based on the average number of parcels corresponding to the next period and the average efficiency of the personnel. Similarly, the server will also dynamically determine the number of equipment required for the next period based on the average number of parcels corresponding to the next period and the average efficiency of the equipment. Average personnel efficiency refers to the average number of parcels that a single staff member can sort per unit time. Average equipment efficiency refers to the average number of parcels that a single working device can sort per unit time.

[0106] In one embodiment, when it is determined that the incoming items in the next time period are abnormal, the warning information triggered and generated by the server will also include the number of personnel and equipment in the current time period, as well as the number of personnel and equipment required in the next time period, so that the management personnel can promptly deploy the staff and work equipment in the next time period based on the warning information.

[0107] In the above embodiment, when it is determined that the incoming parcels in the next time period are abnormal, the staff and work equipment in the next time period are dynamically deployed based on the number of personnel and the number of equipment in the current time period, as well as the parcel quantity information of the next time period, so as to realize the timely allocation of staff and work equipment, and avoid the situation where the number of arranged staff and the number of work equipment do not match the amount of goods to be sorted, such as arranging too many personnel and / or equipment when the amount of goods to be sorted is small, and arranging too few personnel and / or equipment when the amount of goods to be sorted is large, thereby avoiding the accumulation of goods and improving the sorting efficiency.

[0108] In one embodiment, there are multiple current video frames; the multiple video frames are synchronously acquired from corresponding video streams; the multiple video streams are synchronously acquired by corresponding video acquisition devices; and the process starts from step 104 in parallel based on the multiple current video frames.

[0109] Specifically, the logistics transfer site is deployed with multiple video acquisition devices, and the multiple video acquisition devices synchronously capture video streams. The server receives the video streams synchronously captured and sent by the multiple video acquisition devices, synchronously extracts video frames from the multiple synchronously received video streams as current video frames to obtain multiple current video frames, and based on each current video frame in the multiple current video frames, respectively executes the relevant processes in the logistics transfer site early warning management method provided by this application. For each current video frame in the multiple current video frames, the server performs target detection on the current video frame to obtain the sorting status of each cargo in the current video frame, and counts the cargo with a normal sorting status in the current video frame to obtain the total quantity corresponding to the current video frame. According to the total quantity corresponding to each video frame in the current time period in which the current video frame is located, the quantity information of the current time period is determined, and the quantity information of the next time period is predicted based on the quantity information of the current time period. When it is determined that the incoming cargo in the next time period is abnormal based on the quantity information of the current time period and the quantity information of the next time period, an early warning information is triggered.

[0110] It can be understood that the above-mentioned process executed for each current video frame in the multiple current video frames is executed in parallel by the server. In this way, based on the current video frame corresponding to each video acquisition device, the incoming parcels in the next time period in the logistics transfer area covered by the field of view of the corresponding video acquisition device are determined in parallel. When the incoming parcels in the next time period in a certain logistics transfer area are abnormal, the corresponding early warning information is triggered for the logistics transfer area, so that the management personnel can deploy the staff and work equipment in the logistics transfer area in a timely manner based on the early warning information, which can improve the management efficiency of the personnel and equipment in the logistics transfer area, thereby improving the efficiency of cargo sorting.

[0111] In one embodiment, a video capture device is deployed in the logistics transfer yard for each logistics sorting stage. Each video capture device's field of view covers the logistics transfer yard area required for the corresponding logistics sorting stage. Each video capture device deployed in the logistics transfer yard is constantly triggered to capture real-time video streams of the corresponding logistics transfer yard area.

[0112] In one embodiment, each video capture device within a logistics transfer station is equipped with a corresponding early warning device. This early warning device can be integrated into the video capture device as a component, or deployed as a standalone device next to the video capture device. For example, the early warning device can be a buzzer for early warning. Thus, when a warning message is triggered based on the current video frame corresponding to a video capture device, an early warning instruction is sent to the early warning device corresponding to that video capture device, instructing it to issue an early warning.

[0113] In the above embodiment, multiple video capture devices deployed in the logistics transfer site synchronously capture video streams, so that it is convenient to synchronously judge whether the logistics transfer site area corresponding to the corresponding video capture device needs to trigger early warning information based on the synchronously captured video streams, so that management personnel can timely deploy staff and work equipment in the corresponding logistics transfer site area based on the early warning information corresponding to each logistics transfer site area, thereby improving the efficiency of cargo sorting.

[0114] Figure 2 The figure is a flow chart of a logistics transfer site early warning management method in one embodiment, which specifically includes the following steps:

[0115] Step 202: Get the current video frame.

[0116] In step 204, the current video frame is input into the trained target detection model for target detection, and the sorting status and target detection frame of each item in the current video frame are obtained.

[0117] Step 206 : Count the goods in the normal sorting state in the current video frame to obtain the total quantity of goods corresponding to the current video frame.

[0118] Step 208: Determine the piece quantity information of the current time period according to the total piece quantity corresponding to each video frame in the current time period.

[0119] In step 210 , the quantity information of the current period is input into the trained quantity prediction model for prediction to obtain the quantity information of the next period.

[0120] Step 212: When the incoming parcels in the next period are determined to be abnormal based on the parcel volume information of the current period and the parcel volume information of the next period, an early warning message is triggered, and the number of personnel and equipment required in the next period is dynamically determined based on the parcel volume information of the next period.

[0121] Step 214 , dynamically deploying staff for the next period based on the number of staff in the current period and the number of staff required for the next period.

[0122] Step 216 : Dynamically allocate working equipment in the next time period according to the number of equipment in the current time period and the number of equipment required in the next time period.

[0123] Step 218: When the current video frame includes goods with an abnormal sorting status, an alarm message is triggered; the alarm message includes the current video frame and a goods detection frame corresponding to the goods with an abnormal sorting status.

[0124] Figure 3 FIG. 1 is a schematic diagram of the structure of a logistics transfer site early warning management system in one embodiment. The logistics transfer site early warning management system is used to implement a logistics transfer site early warning management method. Figure 3 As shown, the logistics transfer site early warning management system includes a video acquisition device, a server, and a terminal for management personnel. The video acquisition device can specifically be a camera or a surveillance camera. The server includes an edge computing unit and an early warning management unit. The edge computing unit is used to perform target detection on the current video frame to obtain the sorting status of each cargo in the current video frame, and is also used to determine the total quantity of the current video frame based on the sorting status of each cargo in the current video frame, and is also used to determine the quantity information of the current period based on the total quantity of each video frame in the current period. The early warning management unit is used to train the quantity prediction model, and is also used to predict the quantity information of the next period based on the quantity information of the current period through the quantity prediction model. It is also used to determine whether the incoming cargo in the next period is abnormal based on the quantity information of the current period and the quantity information of the next period, and trigger early warning information when the incoming cargo in the next period is abnormal. It is also used to trigger an alarm information when there is cargo with an abnormal sorting status in the current video frame.

[0125] The administrator's terminal is used to receive warning and / or alarm information sent by the server, display the warning and / or alarm information, and issue warnings and / or alarms. After obtaining the warning and / or alarm information corresponding to the current video frame through the terminal, the administrator manually processes the information based on the obtained warning and / or alarm information and dynamically allocates staff and work equipment.

[0126] I understand. Figure 3 The number of edge computing units shown is for example only and is not intended to be limiting. When multiple video capture devices are deployed in a logistics transfer station, each current video frame extracted from the video streams captured by each of the multiple video capture devices is processed by an edge computing unit, thereby enabling parallel processing of multiple current video frames.

[0127] Figure 4 FIG. 1 is a schematic diagram showing the principle of a logistics transfer site early warning management method in one embodiment. Figure 4 As shown, the video stream of the current time period includes multiple video frames (n video frames, n can be customized). When the multiple video frames are processed in sequence according to the time sequence, the multiple video frames will be selected as the current video frames in turn, and based on the selected current video frames, they will be processed in turn according to the target detection and piece quantity statistics process in the logistics transfer site early warning management method provided by this application. The target detection and piece quantity statistics process includes: performing real-time detection on the selected current video frame to obtain the sorting status and cargo detection frame of each cargo in the current video frame, and determining the cargo in the current video frame with a sorting status of normal pieces and abnormal pieces according to the sorting status, that is, determining the normal pieces and abnormal pieces in the current video, and performing statistics on the normal pieces in the current video frame to obtain the total piece quantity of the current video frame, and determining the cargo position of the abnormal pieces in the current video frame based on the cargo detection frame. It is worth noting that Figure 4 Taking the nth video frame as the current video frame as an example, the principle of the logistics transfer site early warning management method provided by this application is explained.

[0128] Furthermore, after sequentially determining the total number of packages in each video frame within the current time period and the location of the abnormal package in each video frame, that is, after determining the total number of packages in the current video frame and the location of the abnormal package in the current video frame, the server determines the package quantity information for the current time period based on the total number of packages in each video frame within the current time period ending with the current video frame. Based on the package quantity information for the current time period, the server predicts the package quantity information for the next time period using a pre-established package quantity prediction model. If, based on the package quantity information for the current time period and the package quantity information for the next time period, it is determined that the incoming packages for the next time period are abnormal, an early warning message is triggered. Furthermore, if an abnormal package exists in the current video frame, an alarm message is triggered. The package quantity prediction model is pre-trained based on video streams from historical time periods.

[0129] It should be understood that although Figure 1 and Figure 2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 and Figure 2 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The order of execution of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0130] In one embodiment, Figure 5 As shown, a logistics transfer site early warning management device 500 is provided, including: an acquisition module 501, a detection module 502, a statistics module 503, a prediction module 504 and an early warning module 505, wherein:

[0131] Acquisition module 501, used to acquire the current video frame;

[0132] Detection module 502, configured to perform target detection on the current video frame to obtain the sorting status of each item in the current video frame;

[0133] The statistics module 503 is used to count the goods in the normal sorting state in the current video frame to obtain the total number of pieces corresponding to the current video frame;

[0134] Prediction module 504, configured to determine the number of pieces in the current time period based on the total number of pieces corresponding to each video frame in the current time period, and predict the number of pieces in the next time period based on the number of pieces in the current time period;

[0135] The early warning module 505 is configured to trigger an early warning message when it is determined that the incoming parcels in the next period are abnormal based on the parcel quantity information of the current period and the parcel quantity information of the next period.

[0136] In one embodiment, the detection module 502 is further configured to perform target detection on the current video frame to obtain the sorting status and detection frame of each item in the current video frame. The logistics transfer station early warning management device 500 also includes an alarm module configured to trigger an alarm message when the current video frame includes items with an abnormal sorting status. The alarm message includes the current video frame and the detection frame corresponding to the item with the abnormal sorting status.

[0137] In one embodiment, the detection module 502 is further configured to input the current video frame into a trained object detection model to perform object detection, thereby obtaining the sorting status and the object detection frame of each item in the current video frame.

[0138] In one embodiment, the prediction module 504 is further configured to input the quantity information of the current period into a trained quantity prediction model for prediction to obtain the quantity information of the next period.

[0139] In one embodiment, the logistics transit site early warning management device 500 also includes: a training module; a training module for obtaining the total quantity of pieces corresponding to each video frame in multiple historical time periods; determining the quantity information of the corresponding historical time period based on the total quantity of pieces corresponding to each video frame in each historical time period; for each pair of adjacent historical time periods, using the quantity information of the previous historical time period as input features and the quantity information of the next historical time period as desired output features for model training to obtain a trained quantity prediction model.

[0140] In one embodiment, the early warning module 505 is also used to dynamically determine the number of personnel and equipment required in the next time period based on the quantity information of the current time period and the quantity information of the next time period when it is determined that the incoming parcels in the next time period are abnormal based on the quantity information of the current time period and the quantity information of the next time period; dynamically allocate the staff in the next time period based on the number of personnel in the current time period and the number of personnel required in the next time period; and dynamically allocate the working equipment in the next time period based on the number of equipment in the current time period and the number of equipment required in the next time period.

[0141] In one embodiment, there are multiple current video frames; the multiple video frames are synchronously acquired from corresponding video streams; the multiple video streams are synchronously acquired by corresponding video acquisition devices; for the multiple current video frames acquired synchronously, the detection module 502, the statistics module 503, the prediction module 504 and the warning module 505 are also used to perform corresponding operations in parallel based on each current video frame.

[0142] The specific limitations of the logistics transfer site early warning management device can be found in the limitations of the logistics transfer site early warning management method described above and will not be further elaborated here. Each module in the aforementioned logistics transfer site early warning management device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of these modules can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each of these modules.

[0143] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store the total number of pieces of each video frame in the current time period. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a logistics transfer site early warning management method is implemented.

[0144] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0145] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the following steps when executing the computer program: obtaining a current video frame; performing target detection on the current video frame to obtain the sorting status of each cargo in the current video frame; counting the cargo in the current video frame with a sorting status of normal pieces to obtain the total quantity of pieces corresponding to the current video frame; determining the quantity information of the current time period based on the total quantity of pieces corresponding to each video frame in the current time period, and predicting the quantity information of the next time period based on the quantity information of the current time period; and triggering an early warning message when it is determined that the incoming pieces of the next time period are abnormal based on the quantity information of the current time period and the quantity information of the next time period.

[0146] In one embodiment, when executing the computer program, the processor further implements the following steps: performing target detection on the current video frame to obtain the sorting status and cargo detection frame of each cargo in the current video frame; triggering an alarm message when the current video frame includes cargo with an abnormal sorting status; the alarm message includes the current video frame and the cargo detection frame corresponding to the cargo with an abnormal sorting status.

[0147] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: inputting the current video frame into a trained target detection model for target detection, and obtaining the sorting status and the target detection frame of each cargo in the current video frame.

[0148] In one embodiment, when the processor executes the computer program, the following steps are further implemented: the quantity information of the current time period is input into a trained quantity prediction model for prediction to obtain the quantity information of the next time period.

[0149] In one embodiment, when the processor executes the computer program, it also implements the following steps: obtaining the total quantity of pieces corresponding to each video frame in multiple historical time periods; determining the quantity information of the corresponding historical time period based on the total quantity of pieces corresponding to each video frame in each historical time period; for each pair of adjacent historical time periods, using the quantity information of the previous historical time period as the input feature and the quantity information of the next historical time period as the expected output feature for model training to obtain a trained quantity prediction model.

[0150] In one embodiment, when the processor executes the computer program, the following steps are also implemented: when it is determined that the incoming parcels in the next period are abnormal based on the parcel quantity information of the current period and the parcel quantity information of the next period, the number of personnel and the number of equipment required in the next period are dynamically determined based on the parcel quantity information of the next period; the staff in the next period are dynamically deployed based on the number of personnel in the current period and the number of personnel required in the next period; the working equipment in the next period is dynamically deployed based on the number of equipment in the current period and the number of equipment required in the next period.

[0151] In one embodiment, there are multiple current video frames; the multiple video frames are synchronously acquired from corresponding video streams; the multiple video streams are synchronously acquired by corresponding video acquisition devices; and when the processor executes the computer program, the following steps are further implemented: starting from the step of performing target detection on the current video frame in parallel based on the multiple current video frames to obtain the sorting status of each cargo in the current video frame.

[0152] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: obtaining a current video frame; performing target detection on the current video frame to obtain the sorting status of each item in the current video frame; counting the items in the current video frame with a normal sorting status to obtain the total quantity of items corresponding to the current video frame; determining the quantity information of the current time period based on the total quantity of items corresponding to each video frame in the current time period, and predicting the quantity information of the next time period based on the quantity information of the current time period; when it is determined that the incoming items of the next time period are abnormal based on the quantity information of the current time period and the quantity information of the next time period, triggering an early warning message.

[0153] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: performing target detection on the current video frame to obtain the sorting status and cargo detection frame of each cargo in the current video frame; when the current video frame includes cargo with an abnormal sorting status, triggering an alarm message; the alarm message includes the current video frame and the cargo detection frame corresponding to the cargo with an abnormal sorting status.

[0154] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: inputting the current video frame into a trained target detection model for target detection, and obtaining the sorting status and the target detection frame of each item in the current video frame.

[0155] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: the quantity information of the current time period is input into a trained quantity prediction model for prediction to obtain the quantity information of the next time period.

[0156] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining the total quantity of each video frame in multiple historical time periods; determining the quantity information of the corresponding historical time period based on the total quantity of each video frame in each historical time period; for each pair of adjacent historical time periods, using the quantity information of the previous historical time period as the input feature and the quantity information of the next historical time period as the expected output feature for model training to obtain a trained quantity prediction model.

[0157] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: when it is determined that the incoming parcels in the next period are abnormal based on the parcel quantity information of the current period and the parcel quantity information of the next period, the number of personnel and the number of equipment required in the next period are dynamically determined based on the parcel quantity information of the next period; the staff in the next period are dynamically deployed based on the number of personnel in the current period and the number of personnel required in the next period; the working equipment in the next period is dynamically deployed based on the number of equipment in the current period and the number of equipment required in the next period.

[0158] In one embodiment, there are multiple current video frames; the multiple video frames are synchronously acquired from corresponding video streams; the multiple video streams are synchronously acquired by corresponding video acquisition devices; and when the computer program is executed by the processor, the following steps are also implemented: starting from the step of performing target detection on the current video frame in parallel based on the multiple current video frames to obtain the sorting status of each item in the current video frame.

[0159] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0160] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0161] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A logistics transfer site early warning management method, characterized in that: The method comprises: Get the current video frame; Performing target detection on the current video frame to obtain a sorting status of each item in the current video frame; Counting the goods in the normal sorting state in the current video frame to obtain a total quantity corresponding to the current video frame; Determining the quantity information of the current time period based on the total quantity corresponding to each video frame in the current time period, and predicting the quantity information of the next time period based on the quantity information of the current time period, wherein the quantity information of the next time period is predicted by a trained quantity prediction model, and the quantity prediction model is trained based on the quantity information of multiple pairs of adjacent historical time periods, where two adjacent historical time periods constitute a pair of adjacent historical time periods; When it is determined based on the package quantity information of the current period and the package quantity information of the next period that the incoming packages of the next period are abnormal, an early warning message is triggered.

2. The method according to claim 1, characterized in that The performing target detection on the current video frame to obtain the sorting status of each cargo in the current video frame includes: Performing target detection on the current video frame to obtain a sorting status and a cargo detection frame of each cargo in the current video frame; The method further comprises: When the current video frame includes goods with an abnormal sorting status, an alarm message is triggered; the alarm message includes the current video frame and a goods detection frame corresponding to the goods with an abnormal sorting status.

3. The method according to claim 2, characterized in that The performing target detection on the current video frame to obtain the sorting status and the cargo detection frame of each cargo in the current video frame includes: The current video frame is input into a trained target detection model for target detection to obtain the sorting status and the target detection frame of each item in the current video frame.

4. The method according to claim 1, wherein The training steps of the quantity prediction model include: Get the total number of pieces corresponding to each video frame in multiple historical periods; Determine the quantity information of the corresponding historical period according to the total quantity corresponding to each video frame in each historical period; For each pair of adjacent historical periods, the quantity information of the previous historical period is used as the input feature, and the quantity information of the next historical period is used as the expected output feature for model training to obtain a trained quantity prediction model.

5. The method according to any one of claims 1 to 4, characterized in that The method further comprises: When it is determined that incoming parcels in the next period are abnormal based on the parcel volume information of the current period and the parcel volume information of the next period, the number of personnel and equipment required in the next period are dynamically determined based on the parcel volume information of the next period; Dynamically deploying staff for the next time period based on the number of staff in the current time period and the number of staff required for the next time period; Dynamically allocate working equipment in the next time period according to the number of equipment in the current time period and the number of equipment required in the next time period.

6. The method according to claim 5, characterized in that There are multiple current video frames; the multiple video frames are synchronously acquired from corresponding video streams; the multiple video streams are synchronously acquired by corresponding video acquisition devices; and execution is started from the step of performing target detection on the current video frame to obtain the sorting status of each cargo in the current video frame based on the multiple current video frames in parallel.

7. A logistics transfer site early warning management device, characterized in that: The device comprises: Acquisition module, used to obtain the current video frame; a detection module, configured to perform target detection on the current video frame to obtain a sorting status of each item in the current video frame; A statistics module, configured to count the goods in the normal sorting state in the current video frame to obtain a total quantity of goods corresponding to the current video frame; a prediction module, configured to determine the quantity information of a current period based on the total quantity corresponding to each video frame in the current period, and predict the quantity information of a next period based on the quantity information of the current period, wherein the quantity information of the next period is predicted by a trained quantity prediction model, and the quantity prediction model is trained based on the quantity information of multiple pairs of adjacent historical periods, where two adjacent historical periods constitute a pair of adjacent historical periods; The early warning module is used to trigger an early warning message when it is determined that the incoming parcels in the next time period are abnormal based on the parcel quantity information of the current time period and the parcel quantity information of the next time period.

8. The device according to claim 7, wherein The detection module is further configured to perform target detection on the current video frame to obtain a sorting status and a cargo detection frame for each cargo in the current video frame. The device further comprises an alarm module configured to trigger an alarm message when the current video frame includes cargo with an abnormal sorting status. The alarm message includes the current video frame and the cargo detection frame corresponding to the cargo with an abnormal sorting status.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images