Pass rate prediction model training method, image acquisition device parameter value determination method

Abstract

The application discloses a qualified rate prediction model training method and an image acquisition equipment parameter value determination method. When the parameter values of the resolution, exposure rate, brightness and direction value and the like of an existing image acquisition equipment are not properly set, the image acquisition equipment may miss some target images during snapshotting. Since the qualified rate prediction model in the method provided by the application is trained in advance based on the corresponding relationship between each first parameter value set and the qualified rate, the parameter value of each to-be-adjusted parameter value of the image acquisition equipment can be qualified rate predicted through the qualified rate prediction model, so that the parameter value of the to-be-adjusted parameter with the highest predicted qualified rate is determined and is used as the finally determined target parameter value, thereby solving the problem that the image acquisition equipment misses some target images during snapshotting due to improper parameter value setting, and improving the snapshot qualified rate of the image acquisition equipment.

Description

Technical Field

[0001] The present invention relates to the field of video surveillance technology, and in particular to a method for training a pass rate prediction model and a method for determining parameter values ​​of an image acquisition device. Background Art

[0002] In recent years, with the rapid development of computer applications, video surveillance has become widely used in everyday scenarios, including vehicle capture and facial recognition. By analyzing target images such as license plates and faces captured by image acquisition devices, abnormal events can be promptly addressed. Therefore, for video surveillance, capturing the target image is particularly important.

[0003] In order to capture more target images, the capture pass rate of the image acquisition device must be improved first, where the capture pass rate refers to the ratio of the number of target images actually captured to the number of target images expected to be captured, and the number of target images expected to be captured refers to the number of target images obtained by analyzing the video data stream captured by the image acquisition device. There are two main methods for improving the capture pass rate of image acquisition devices in the prior art. One is to change the hardware of the image acquisition device, and the other is to post-process the captured data through software.

[0004] In an existing method for collecting facial information based on camera face recognition, a main controller collects video data and caches it in a main memory. The main controller then transmits the segmented video data from the main memory to an image processing chip. The main controller then receives facial information and facial image quality parameters from the segmented video data analyzed by the image processing chip. The main controller then captures a facial photo based on the facial information and facial image quality parameters meeting preset standards, stores it, and uploads it. However, when the data volume in this existing technology is large, the image processing chip's performance is limited, which reduces the image acquisition device's pass rate and processing efficiency.

[0005] Therefore, due to the defects of the above-mentioned existing technology, the existing technology will affect the original performance of the image acquisition device. How to improve the capture pass rate of the image acquisition device without changing the hardware structure and original performance of the image acquisition device becomes a technical problem that needs to be solved urgently. Summary of the Invention

[0006] The present invention provides a method for training a qualified rate prediction model, a method, an apparatus, a device and a medium for determining parameter values ​​of an image acquisition device, so as to solve the problem of low qualified rate of snapshots of image acquisition devices in the prior art.

[0007] The present invention provides a method for training a pass rate prediction model, the method comprising:

[0008] According to a sample set consisting of a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and a qualified rate, obtaining any first parameter value set and a corresponding qualified rate in the sample set;

[0009] Inputting each first parameter value in the first parameter value set into the original pass rate prediction model, and obtaining the output predicted pass rate corresponding to the first parameter value set;

[0010] According to the pass rate and predicted pass rate corresponding to the first parameter value set, the parameter values ​​of each parameter of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

[0011] Furthermore, the process of determining the corresponding relationship includes:

[0012] For each preset first parameter value set of each parameter to be adjusted, sending each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value;

[0013] Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number;

[0014] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

[0015] Furthermore, after obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model, the method further includes:

[0016] Normalize each first parameter value in the first parameter value set.

[0017] Accordingly, the present invention provides a method for determining parameter values ​​of an image acquisition device, the method comprising:

[0018] Determining a target first parameter value set corresponding to a target pass rate with the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0019] Fine-tune the first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value to obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and combine the first parameter value and the second parameter value corresponding to each parameter to be adjusted to obtain each second parameter value set other than the target first parameter value set;

[0020] Based on the pre-trained pass rate prediction model, each predicted pass rate corresponding to each input second parameter value set is determined, and according to each predicted pass rate and the target pass rate, the parameter value in the parameter value set corresponding to the highest pass rate is determined as the target parameter value of the image acquisition device.

[0021] Furthermore, determining, based on each predicted qualified rate and the target qualified rate, a parameter value in a parameter value set corresponding to a highest qualified rate as a target parameter value of the image acquisition device includes:

[0022] If the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate, performing at least one fine adjustment based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine adjustment value, and determining the target parameter value of the image acquisition device based on whether the target predicted qualified rate determined after two adjacent fine adjustments is less than the target predicted qualified rate of the previous one;

[0023] If the target predicted qualified rate is not greater than the target qualified rate, the first parameter value in the target first parameter value set is determined as the target parameter value of the image acquisition device.

[0024] Furthermore, determining the target parameter value of the image acquisition device based on whether the target prediction pass rate determined after two adjacent fine-tunings is less than the target prediction pass rate of the previous one includes:

[0025] Based on the pass rate prediction model, the second parameter value after the last fine-tuning is predicted. If the predicted target prediction pass rate for the next time is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the target prediction pass rate for the next time, and the fine-tuning process and the prediction process are performed again until the predicted target prediction pass rate for the next time is less than the predicted target prediction pass rate for the previous time, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

[0026] Furthermore, after obtaining each second parameter value set by combining each parameter value corresponding to each parameter to be adjusted, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model, the method further includes:

[0027] Each second parameter value in each second parameter value set is normalized.

[0028] Accordingly, the present invention provides a qualified rate prediction model training device, the device comprising:

[0029] an acquisition module, configured to acquire any first parameter value set and the corresponding pass rate in a sample set consisting of a predetermined correspondence relationship between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0030] A training module is used to input each first parameter value in the first parameter value set into the original pass rate prediction model to obtain the predicted pass rate corresponding to the output first parameter value set; according to the pass rate and predicted pass rate corresponding to the first parameter value set, the parameter values ​​of each parameter of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

[0031] Furthermore, the device further comprises:

[0032] a determination module, configured to, for each first parameter value set of each parameter to be adjusted, send each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value;

[0033] Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number;

[0034] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

[0035] Furthermore, the device further comprises:

[0036] A processing module is used to normalize each first parameter value in the first parameter value set after obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model.

[0037] Accordingly, the present invention provides a device for determining parameter values ​​of an image acquisition device, the device comprising:

[0038] a determination module, configured to determine a target first parameter value set corresponding to a target pass rate having the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0039] a fine-tuning module, configured to fine-tune a first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value, obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and obtain each second parameter value set other than the target first parameter value set by combining the first parameter value and the second parameter value corresponding to each parameter to be adjusted;

[0040] A prediction module, configured to determine each predicted pass rate corresponding to each input second parameter value set based on a pre-trained pass rate prediction model;

[0041] The determination module is further configured to determine, based on each predicted qualified rate and the target qualified rate, a parameter value in a parameter value set corresponding to a highest qualified rate as a target parameter value of the image acquisition device.

[0042] Furthermore, the determination module is specifically used to perform at least one fine-tuning based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine-tuning value if the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate; and determine the target parameter value of the image acquisition device based on whether the target predicted qualified rate of the latter one is less than the target predicted qualified rate of the previous one among the target predicted qualified rates determined after two adjacent fine-tunings; if the target predicted qualified rate is not greater than the target qualified rate, determine the first parameter value in the target first parameter value set as the target parameter value of the image acquisition device.

[0043] Furthermore, the determination module is specifically used to predict the second parameter value after the last fine-tuning based on the pass rate prediction model. If the predicted next target prediction pass rate is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the next target prediction pass rate and the fine-tuning process and prediction process are performed again until the predicted next target prediction pass rate is less than the predicted previous target prediction pass rate, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

[0044] Furthermore, the device further comprises:

[0045] A processing module is used to normalize each second parameter value in each second parameter value set after combining each parameter value corresponding to each parameter to be adjusted, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model.

[0046] Accordingly, the present invention provides an electronic device, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; a computer program is stored in the memory, and when the program is executed by the processor, the processor executes the computer program stored in the memory to implement the steps of any of the above-mentioned methods for training a pass rate prediction model or any of the above-mentioned methods for determining parameter values ​​of an image acquisition device.

[0047] Accordingly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned methods for training a pass rate prediction model or any of the above-mentioned methods for determining parameter values ​​of an image acquisition device.

[0048] The present invention provides a method for training a pass rate prediction model, a method, apparatus, equipment and medium for determining parameter values ​​of an image acquisition device. When the parameter values ​​of parameters such as resolution, exposure rate, brightness and direction value of existing image acquisition devices are improperly set, the image acquisition device may miss part of the target image when capturing. Since the pass rate prediction model in the method provided by the present invention is pre-trained based on the correspondence between each first parameter value set and the pass rate, and since the pass rate prediction model can be used to predict the pass rate of the parameter value of each parameter value to be adjusted of the image acquisition device, the parameter value of the parameter to be adjusted when the predicted pass rate is the highest is determined and used as the final target parameter value, thereby solving the problem that the image acquisition device misses part of the target image when capturing due to improper parameter value setting, and improving the capture pass rate of the image acquisition device. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 A schematic diagram of a process of a pass rate prediction model training method provided by an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of a BP model provided in an embodiment of the present invention;

[0052] Figure 3 A schematic diagram of a process of a pass rate prediction model provided by an embodiment of the present invention;

[0053] Figure 4 A schematic diagram of a method for determining parameter values ​​of an image acquisition device provided by an embodiment of the present invention;

[0054] Figure 5 A schematic diagram of the structure of a pass rate prediction model training device provided by an embodiment of the present invention;

[0055] Figure 6 A schematic structural diagram of a device for determining parameter values ​​of an image acquisition device provided by an embodiment of the present invention;

[0056] Figure 7 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention;

[0057] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some, not all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0059] In order to improve the capture pass rate of image acquisition equipment, embodiments of the present invention provide a pass rate prediction model training method, a parameter value determination method, an apparatus, a device and a medium for image acquisition equipment.

[0060] Example 1:

[0061] Figure 1 A schematic diagram of a process of a pass rate prediction model training method provided by an embodiment of the present invention, the process includes the following steps:

[0062] S101: According to a predetermined sample set consisting of a correspondence between each first parameter value set of each parameter to be adjusted of an image acquisition device and a qualified rate, obtaining any first parameter value set and a corresponding qualified rate in the sample set.

[0063] The embodiments of the present invention provide a yield prediction model for an electronic device, wherein the electronic device can be a smart terminal such as a PC, tablet computer, or mobile terminal, or a server; the server can be a local server or a cloud server. Specifically, the embodiments of the present invention do not impose any restrictions on this.

[0064] In an embodiment of the present invention, in order to improve the capture pass rate of the image acquisition device, the pass rate prediction model is first trained. In the embodiment of the present invention, a sample set for training is saved, and the sample set includes the correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate.

[0065] The first parameter value set includes the first parameter value of each parameter value to be adjusted of the image acquisition device, each parameter to be adjusted includes resolution, exposure rate, brightness value and direction value, etc., and each first parameter value set has a one-to-one relationship with the pass rate.

[0066] Obtain any first parameter value set and the corresponding pass rate in the sample set. The pass rate corresponding to the first parameter value set is predetermined, indicating the pass rate of capturing when the parameter to be adjusted of the image acquisition device is the first parameter value in the first parameter value set.

[0067] S102: Input each first parameter value in the first parameter value set into an original pass rate prediction model, and obtain the output predicted pass rate corresponding to the first parameter value set.

[0068] After obtaining any parameter value set and the corresponding pass rate in the sample set, each first parameter value in the first parameter value set is input into the original pass rate prediction model, and the original pass rate prediction model outputs the predicted pass rate corresponding to the first parameter value set, where the predicted pass rate identifies the pass rate of the image acquisition device predicted by the pass rate prediction model.

[0069] The original pass rate prediction model is a convolutional neural network model using a deep learning method, or it can be a multi-layer feedforward network (BP) model trained by an error back propagation algorithm. Specifically, the embodiment of the present invention does not impose any restrictions on this.

[0070] For example, when the original pass rate prediction model is the BP model, Figure 2 A schematic diagram of a BP model provided by an embodiment of the present invention is shown as follows: Figure 2 As shown, each first parameter value in the first parameter value set is input to each node of the input layer of the BP model, and the input layer performs feature extraction on each first parameter value to determine the input layer vector corresponding to each node of the input layer. The number of nodes included in the input layer is the same as the number of first parameter values ​​included in the first parameter value set.

[0071] According to the input layer vector and the pre-saved weight connection matrix between the input layer and the hidden layer, the activation function is used to process the hidden layer to determine the hidden layer vector output by the hidden layer; according to the hidden layer vector and the pre-saved weight connection matrix between the hidden layer and the output layer, the activation function is used to process the output layer to determine the prediction qualification rate of the output layer output.

[0072] S103: According to the pass rate and the predicted pass rate corresponding to the first parameter value set, the parameter values ​​of the parameters of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

[0073] After determining the predicted pass rate corresponding to the first parameter value set output by the original pass rate prediction model, the original pass rate prediction model is trained according to the predicted pass rate and the pass rate to adjust the parameter values ​​of various parameters of the original pass rate prediction model.

[0074] The above operation is performed on each first parameter value set included in the sample set used to train the original pass rate prediction model. When a preset condition is met, a trained pass rate prediction model is obtained. The preset condition may be that the sum of the errors obtained after training the original pass rate prediction model for the first parameter values ​​in each first parameter value set in the sample set is less than a preset error threshold, or that the number of first parameter value sets whose predicted pass rate is consistent with the pass rate is greater than a set number; or that the number of iterations of training the original pass rate prediction model reaches a set maximum number of iterations, etc. Specifically, this application does not impose any restrictions on this.

[0075] As a possible implementation method, when training the original pass rate prediction model, the first parameter value set in the sample set can be divided into a first training parameter value set and a first test parameter value set. The original pass rate prediction model is first trained based on the first training parameter value set, and then the reliability of the trained pass rate prediction model is tested based on the first test parameter value set.

[0076] In the embodiment of the present invention, when the parameter values ​​of parameters such as resolution, exposure rate, brightness and direction value of the existing image acquisition device are improperly set, the image acquisition device may miss part of the target image when capturing. Since the qualified rate prediction model in the method provided by the present invention is pre-trained based on the correspondence between each first parameter value set and the qualified rate, and since the qualified rate prediction model can be used to predict the qualified rate of each parameter value to be adjusted of the image acquisition device, the parameter value of the parameter to be adjusted when the predicted qualified rate is the highest is determined and used as the final target parameter value, thereby solving the problem of the image acquisition device missing part of the target image when capturing due to improper parameter value setting, and improving the captured qualified rate of the image acquisition device.

[0077] Example 2:

[0078] In order to determine the corresponding relationship between each first parameter value set and the pass rate, based on the above embodiment, in an embodiment of the present invention, the process of determining the corresponding relationship includes:

[0079] For each preset first parameter value set of each parameter to be adjusted, sending each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value;

[0080] Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number;

[0081] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

[0082] In order to determine the correspondence between each first parameter value set and the pass rate, in an embodiment of the present invention, for each parameter to be adjusted of the image acquisition device, each first parameter value set preset by the user is saved. For each first parameter value set, each first parameter value in the first parameter value set is sent to the image acquisition device, so that the parameter value of each parameter to be adjusted of the image acquisition device is the corresponding first parameter value.

[0083] After the parameter value of each parameter to be adjusted of the image acquisition device is set to the corresponding first parameter value, the video data within the preset time length is recognized and captured in real time. Specifically, the recognition is based on a pre-saved preset algorithm to determine each first target image to be captured and the first number of first target images.

[0084] The first target image is an image of a target object that meets the preset image conditions and is captured by the image acquisition device. The target object includes at least one of a face image and a vehicle license plate. When the target object is a face image, the preset algorithm is a face recognition algorithm. The preset image is required to include the entire face area and the image clarity is higher than the preset clarity threshold; when the target object is a vehicle license plate, the preset algorithm is a license plate recognition algorithm. The preset image is required to include the entire license plate area and the vehicle's numbers and letters can be successfully recognized.

[0085] The image acquisition device sends the first number to the electronic device. The electronic device receives the first number and determines, based on the first number and a pre-stored second number, that the ratio of the first number to the second number is the qualified rate corresponding to the first parameter value.

[0086] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established, and the determined corresponding relationship is saved.

[0087] Example 3:

[0088] In order to implement the training of the pass rate prediction model, based on the above embodiments, in an embodiment of the present invention, after obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model, the method further includes:

[0089] Normalize each first parameter value in the first parameter value set.

[0090] In an embodiment of the present invention, when the original pass rate prediction model is a BP model, since the activation function in the BP model requires the input and output layers to be within the range of -1 and 1, the embodiment of the present invention, after obtaining any first parameter value set and the corresponding pass rate in the sample set, also normalizes each first parameter value in the first parameter value set, and inputs each first parameter value after normalization into the original pass rate prediction model for training.

[0091] The normalized data provided by the embodiment of the present invention is shown in Table 1:

[0092]

[0093] Table 1

[0094] As shown in Table 1, the normalized value of the resolution value in a first parameter value set of the image acquisition device is 0.7615941559557649, the normalized value of the exposure value is 0.9640275800758169, and the normalized value of the brightness value is 0.9640275800758169. The qualified rate corresponding to this first parameter value set is 0.7384218713128035. The same applies to other first parameter value sets, which will not be described in detail in this embodiment of the present invention.

[0095] Example 4:

[0096] The following describes a specific example of a pass rate prediction model of the present invention. Figure 3 A process diagram of a pass rate prediction model provided by an embodiment of the present invention is shown in FIG. Figure 3 As shown, the process includes the following steps:

[0097] S301: Obtain any first training parameter value set and the corresponding pass rate in the sample set, input each first training parameter value in the first training parameter value set into the original pass rate prediction model, and obtain the predicted pass rate corresponding to the output first parameter value set.

[0098] S302: Training the original pass rate prediction model according to the pass rate and the predicted pass rate corresponding to the first parameter value set.

[0099] S303: Determine whether the error sum value is less than a preset error threshold, or whether the number of training iterations is greater than the maximum number of iterations. If so, proceed to S304; if not, return to S301.

[0100] S304: Obtaining a trained pass rate prediction model.

[0101] S305: Test the qualified rate prediction model according to the first test parameter value set to determine whether the error sum value is less than a preset error threshold. If so, proceed to S306; if not, return to S301.

[0102] S306: Obtain a revised pass rate prediction model.

[0103] Example 5:

[0104] The following describes a method for training a pass rate prediction model of the present invention through a specific embodiment. When the pass rate prediction model is a BP model, a data set consisting of the corresponding relationship between each first parameter value set of each parameter to be adjusted of a predetermined image acquisition device and the pass rate is passed into the BP model established by the BP algorithm.

[0105] The input layer vector of the BP model is represented by x, where x = {x1, x2, x3, ..., x i ,…,x n}, where i represents the parameter value of the i-th parameter to be adjusted, and the hidden layer vector is represented by h, h={h1, h2, h3, ..., h i ,…,h n}, the output layer vector is represented by y, where y = {y1}, y1 represents the output predicted pass rate, the expected output vector is represented by d, where d = {d1}, d1 represents the pass rate corresponding to the first parameter value set; the weight connection matrix between the input layer and the hidden layer is represented by v; the weight connection matrix between the hidden layer and the output layer is represented by w.

[0106] The number of input layer nodes of the BP model is generally determined by trial and error, and the formula is as follows: Where m is the number of nodes in the hidden layer to be estimated, n is the number of nodes in the input layer, which is 4, l is the number of nodes in the output layer, and through trial calculation, the relatively optimal number of nodes in the hidden layer is 2. v is a matrix with 2 rows and 4 columns, and w is a matrix with 2 rows and 1 column.

[0107] The maximum number of iterations of the BP model is preset to M, the error threshold is preset to ∈, and the activation function is the unipolar S-shaped growth curve (sigmoid) function: f(x) = 1 / (1+e-x ).

[0108] Select any first parameter value set X = {X1, X2, X3, X4} and get the hidden layer vector h j and the output layer vector y k ,in For example, when j=1, Each element of the first row of the matrix v is multiplied by each parameter value in the sample X, the sum of the product values ​​is calculated, and the sum is substituted into the sigmoid function as the independent variable x to obtain h1; when j = 2, Each element of the second row of the matrix v is multiplied by each parameter value in the sample X to obtain the sum of the products, and the sum is substituted into the sigmoid function as the independent variable x to obtain h2.

[0109] in, k=1, each element of the first column of matrix w is the same as the hidden layer vector h j Multiply each vector value of , calculate the sum of the product values, and substitute the sum as the independent variable x into the sigmoid function to obtain the output layer vector y1.

[0110] y composed of the output layer vector y1 corresponding to each first parameter value set X k and the expected output value d k , the error function of each round of training is determined as

[0111] Determine the partial derivative of the error function with respect to each neuron in the output layer, where the vector of each neuron in the output layer is y o , The error function E is applied to the vector y of each neuron in the output layer o The partial derivative of Specifically, Determine the partial derivative of the error function E with respect to each neuron in the hidden layer, where the vector of each neuron in the hidden layer is h o , The error function E is the vector h of each neuron in the hidden layer o The partial derivative of Specifically,

[0112] According to the partial derivatives of each round of training and After each round of training, the connection weights of the BP model are adjusted. The adjusted connection weights from the hidden layer to the output layer are The connection weight from the input layer to the hidden layer is in in Refers to the set learning rate constant.

[0113] Calculate the global error and value E, If the global error and value E are less than the preset error threshold ∈, or the number of training times reaches the maximum number of iterations M, the trained BP model is obtained.

[0114] Example 6:

[0115] Figure 4 A schematic diagram of a method for determining parameter values ​​of an image acquisition device provided by an embodiment of the present invention includes the following steps:

[0116] S401: Determine a target first parameter value set corresponding to a target pass rate with the highest pass rate according to a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate.

[0117] An embodiment of the present invention provides a method for determining parameter values ​​of an image acquisition device, which is applied to an electronic device. The electronic device can be a smart terminal such as a PC, tablet computer, or mobile terminal, or a server; the server can be a local server or a cloud server. The electronic device is the same device as the electronic device that performs the yield prediction model training method.

[0118] Based on the predetermined correspondence between each first parameter value set of each parameter to be adjusted for the image acquisition device and the pass rate, a target pass rate with the highest pass rate is determined within the correspondence, and a target first parameter value set corresponding to the target pass rate is determined. The first parameter value in the target first parameter value set is the roughly determined parameter value of the image acquisition device.

[0119] S402: Fine-tune the first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value to obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and combine the first parameter value and the second parameter value corresponding to each parameter to be adjusted to obtain each second parameter value set except the target first parameter value set.

[0120] In order to improve the capture pass rate of the image acquisition device, in an embodiment of the present invention, the first parameter value in the target first parameter value set is also fine-tuned, and the electronic device pre-saves it as a fine-tuning value, wherein the fine-tuning value is preset. Preferably, the fine-tuning value can be a value such as 0.1 or 0.2.

[0121] Based on the preset fine-tuning value, the first parameter value of each parameter to be adjusted in the target first parameter value set is fine-tuned, and the preset fine-tuning value is continuously added and subtracted on the basis of the first parameter value to obtain a set number of second parameter values ​​after fine-tuning on both sides of the adjacent first parameter value.

[0122] For example, when the set quantity is 2, the first parameter value is 9, and the preset fine-tuning value is 0.1, adding 0.1 to 9 gets 9.1, and subtracting 0.1 from 9 gets 8.9, resulting in two second parameter values ​​of 8.9 and 9.1; when the set quantity is 4, the first parameter value is 9, and the preset fine-tuning value is 0.1, adding 0.1 to 9 gets 9.1, adding 0.1 to 9.1 gets 9.2, subtracting 0.1 from 9 gets 8.9, and subtracting 0.1 from 8.9 gets 8.8.

[0123] According to the first parameter value and the second parameter value corresponding to each parameter to be adjusted, the parameter values ​​corresponding to each parameter value to be adjusted are combined to obtain each second parameter value set except the target first parameter value set.

[0124] For example, each parameter to be adjusted includes image acquisition device parameter 1, image acquisition device parameter 2, and image acquisition device parameter 3; the first parameter value corresponding to image acquisition device parameter 1 is 9, and the corresponding second parameter values ​​include 8.9 and 9.1; the first parameter value corresponding to image acquisition device parameter 2 is 10.2, and the corresponding second parameter values ​​include 10.1 and 10.3; the first parameter value corresponding to image acquisition device parameter 3 is 14.1, and the corresponding second parameter values ​​include 14 and 14.2.

[0125] The parameter values ​​corresponding to image acquisition device parameter 1, image acquisition device parameter 2, and image acquisition device parameter 3 are combined to obtain a parameter value set as shown in Table 2:

[0126] Image acquisition device parameters 1 Image acquisition device parameters 2 Image acquisition device parameters 3 8.9 10.1 14 8.9 10.1 14.1 8.9 10.1 14.2 9 10.2 14 9 10.2 14.1 9 10.2 14.2 9.1 10.3 14 9.1 10.3 14.1 9.1 10.3 14.2

[0127] Table 2

[0128] The parameter value set corresponding to the second row in Table 2 includes: parameter value 8.9 for image acquisition device parameter 1, parameter value 10.1 for image acquisition device parameter 2, and parameter value 14 for image acquisition device parameter 3; the parameter value set corresponding to the third row includes: parameter value 8.9 for image acquisition device parameter 1, parameter value 10.1 for image acquisition device parameter 2, and parameter value 14.1 for image acquisition device parameter 3. The parameter value sets corresponding to other rows in Table 1 are similar and are not further described in detail in this embodiment of the present invention.

[0129] The parameter value set corresponding to the sixth row in Table 2 is the target first parameter value set. After removing the target first parameter value set from the parameter value set corresponding to each row in Table 1, each second parameter value set is obtained.

[0130] S403: Based on the pre-trained pass rate prediction model, determine each predicted pass rate corresponding to each second parameter value set input, and according to each predicted pass rate and the target pass rate, determine the parameter value in the parameter value set corresponding to the highest pass rate as the target parameter value of the image acquisition device.

[0131] In order to determine the target parameter value of the image acquisition device so as to improve the pass rate of the image acquisition device, a pass rate prediction model pre-trained according to the method of the above embodiment is used, and each parameter value in each second parameter value set is input into the pass rate prediction model to determine each predicted pass rate corresponding to each second parameter value set output by the pass rate pre-model.

[0132] Based on each predicted pass rate corresponding to each second parameter value set and the target pass rate corresponding to the target first parameter value set, the highest pass rate can be determined. The highest pass rate may be the predicted pass rate or the target pass rate. The parameter value in the parameter value set corresponding to the highest pass rate is determined as the target parameter value of the image acquisition device.

[0133] Example 7:

[0134] To determine the target parameter value of the image acquisition device, based on the above embodiment, in an embodiment of the present invention, determining, based on each predicted pass rate and the target pass rate, that the parameter value in the parameter value set corresponding to the highest pass rate is the target parameter value of the image acquisition device includes:

[0135] If the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate, performing at least one fine adjustment based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine adjustment value, and determining the target parameter value of the image acquisition device based on whether the target predicted qualified rate determined after two adjacent fine adjustments is less than the target predicted qualified rate of the previous one;

[0136] If the target predicted qualified rate is not greater than the target qualified rate, the first parameter value in the target first parameter value set is determined as the target parameter value of the image acquisition device.

[0137] In order to determine the target parameter value of the image acquisition device, in an embodiment of the present invention, based on each predicted pass rate output by the pass rate prediction model, the highest target predicted pass rate among each predicted pass rate is determined. If the target predicted pass rate is greater than the target pass rate, it means that the parameter value in the target second parameter value set corresponding to the target predicted pass rate is still not the target parameter value with the best pass rate that can enable the image acquisition device to capture the image. Therefore, the second parameter value in the target second parameter value set can also be fine-tuned.

[0138] Specifically, for the second parameter value in the target second parameter value set corresponding to the target prediction pass rate, at least one fine-tuning is performed on the second parameter value in the target second parameter value set corresponding to the preset fine-tuning value. Based on the target prediction pass rate determined after two adjacent fine-tunings, whether the latter target prediction pass rate is less than the previous target prediction pass rate, the target parameter value of the image acquisition device is determined.

[0139] If the target prediction pass rate of the subsequent time is less than the target prediction pass rate of the previous time, the parameter value in the target second parameter value set corresponding to the target prediction pass rate of the previous time is determined as the target parameter value of the image acquisition device; if the target prediction pass rate of the subsequent time is not less than the target prediction pass rate of the previous time, the target prediction pass rate of the subsequent time of the previous time is updated, and the target parameter value of the image acquisition device is determined.

[0140] If the target predicted qualified rate is not greater than the target qualified rate, it is determined that the first parameter value in the target first parameter value set corresponding to the target qualified rate is the target parameter value of the image acquisition device.

[0141] Example 8:

[0142] To determine the target parameter value of the image acquisition device, based on the above embodiments, in an embodiment of the present invention, determining the target parameter value of the image acquisition device according to whether the target predicted pass rate determined after two adjacent fine-tunings is less than the target predicted pass rate of the previous one includes:

[0143] Based on the pass rate prediction model, the second parameter value after the last fine-tuning is predicted. If the predicted target prediction pass rate for the next time is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the target prediction pass rate for the next time, and the fine-tuning process and the prediction process are performed again until the predicted target prediction pass rate for the next time is less than the predicted target prediction pass rate for the previous time, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

[0144] In order to determine the target parameter value of the image acquisition device, when it is determined that the highest target predicted pass rate in each predicted pass rate is greater than the target pass rate, the second parameter value in the second parameter value set after the last fine-tuning is predicted based on the pass rate prediction model to obtain the next target predicted pass rate.

[0145] According to the target prediction pass rate of the next time and the target prediction pass rate of the previous time, if the target prediction pass rate of the next time is not less than the target prediction pass rate of the previous time, the target second parameter value set is updated according to the second parameter value set corresponding to the target prediction pass rate of the next time, and the fine-tuning process and the prediction process are performed again. If the predicted target prediction pass rate of the next time is not less than the target prediction pass rate of the previous time, the target second parameter value set is updated, fine-tuned and predicted in a cycle until the predicted target prediction pass rate of the next time is less than the predicted target prediction pass rate of the previous time.

[0146] Since the target prediction pass rate obtained in the previous prediction is the highest pass rate, the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device. When the parameter value of the image acquisition device is determined to be the target parameter value, the pass rate for capturing the image is the highest.

[0147] Example 9:

[0148] In order to predict each predicted pass rate corresponding to each second parameter value set, based on the above embodiments, in an embodiment of the present invention, after combining each parameter value corresponding to each parameter to be adjusted to obtain each second parameter value set, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model, the method further includes:

[0149] Each second parameter value in each second parameter value set is normalized.

[0150] In order to predict each predicted pass rate corresponding to each second parameter value set, each second parameter value in each second parameter value set must be normalized before inputting it into the pass rate prediction model. The method for normalizing values ​​is known in the art and will not be further described in detail in the present embodiment.

[0151] Example 10:

[0152] Based on the above embodiments, Figure 5 A schematic diagram of a device for training a pass rate prediction model according to an embodiment of the present invention is provided, wherein the device comprises:

[0153] An acquisition module 501 is configured to acquire any first parameter value set and the corresponding pass rate in a sample set consisting of a predetermined correspondence relationship between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0154] The training module 502 is used to input each first parameter value in the first parameter value set into the original pass rate prediction model to obtain the predicted pass rate corresponding to the output first parameter value set; according to the pass rate and predicted pass rate corresponding to the first parameter value set, the parameter values ​​of each parameter of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

[0155] Furthermore, the device further comprises:

[0156] a determination module, configured to, for each first parameter value set of each parameter to be adjusted, send each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value;

[0157] Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number;

[0158] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

[0159] Furthermore, the device further comprises:

[0160] A processing module is used to normalize each first parameter value in the first parameter value set after obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model.

[0161] Example 11:

[0162] Based on the above embodiments, Figure 6 A schematic structural diagram of a device for determining parameter values ​​of an image acquisition device provided by an embodiment of the present invention, the device comprising:

[0163] A determination module 601 is configured to determine a target first parameter value set corresponding to a target pass rate having the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0164] a fine-tuning module 602 configured to fine-tune the first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value, obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and obtain each second parameter value set other than the target first parameter value set by combining the first parameter value and the second parameter value corresponding to each parameter to be adjusted;

[0165] Prediction module 603, configured to determine each predicted pass rate corresponding to each input second parameter value set based on a pre-trained pass rate prediction model;

[0166] The determination module 601 is further configured to determine, based on each predicted qualified rate and the target qualified rate, a parameter value in a parameter value set corresponding to a highest qualified rate as a target parameter value of the image acquisition device.

[0167] Furthermore, the determination module is specifically used to perform at least one fine-tuning based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine-tuning value if the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate; and determine the target parameter value of the image acquisition device based on whether the target predicted qualified rate of the latter one is less than the target predicted qualified rate of the previous one among the target predicted qualified rates determined after two adjacent fine-tunings; if the target predicted qualified rate is not greater than the target qualified rate, determine the first parameter value in the target first parameter value set as the target parameter value of the image acquisition device.

[0168] Furthermore, the determination module is specifically used to predict the second parameter value after the last fine-tuning based on the pass rate prediction model. If the predicted next target prediction pass rate is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the next target prediction pass rate and the fine-tuning process and prediction process are performed again until the predicted next target prediction pass rate is less than the predicted previous target prediction pass rate, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

[0169] Furthermore, the device further comprises:

[0170] A processing module is used to normalize each second parameter value in each second parameter value set after combining each parameter value corresponding to each parameter to be adjusted, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model.

[0171] Example 12:

[0172] Figure 7 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Based on the above embodiments, an embodiment of the present invention further provides an electronic device, such as Figure 7 As shown, it includes: a processor 701 , a communication interface 702 , a memory 703 and a communication bus 704 , wherein the processor 701 , the communication interface 702 and the memory 703 communicate with each other via the communication bus 704 .

[0173] The memory 703 stores a computer program. When the program is executed by the processor 701, the processor 701 performs the following steps:

[0174] According to a sample set consisting of a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and a qualified rate, obtaining any first parameter value set and a corresponding qualified rate in the sample set;

[0175] Inputting each first parameter value in the first parameter value set into the original pass rate prediction model, and obtaining the output predicted pass rate corresponding to the first parameter value set;

[0176] According to the pass rate and predicted pass rate corresponding to the first parameter value set, the parameter values ​​of each parameter of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

[0177] Furthermore, the processor 701 is further configured to:

[0178] For each preset first parameter value set of each parameter to be adjusted, sending each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value;

[0179] Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number;

[0180] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

[0181] Furthermore, the processor 701 is further configured to, after obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model, further comprise:

[0182] Normalize each first parameter value in the first parameter value set.

[0183] The communication bus mentioned in the electronic device mentioned above may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in the figure, but this does not mean that there is only one bus or only one type of bus.

[0184] The communication interface 702 is used for communication between the electronic device and other devices.

[0185] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk memory. Alternatively, the memory may be at least one storage device located away from the processor.

[0186] The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; it can also be a digital signal processing processor (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0187] Example 13:

[0188] Figure 8 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Based on the above embodiments, an embodiment of the present invention further provides an electronic device, such as Figure 8As shown, it includes: a processor 801 , a communication interface 802 , a memory 803 and a communication bus 804 , wherein the processor 801 , the communication interface 802 and the memory 803 communicate with each other via the communication bus 804 .

[0189] The memory 803 stores a computer program. When the program is executed by the processor 801, the processor 801 performs the following steps:

[0190] Determining a target first parameter value set corresponding to a target pass rate with the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0191] Fine-tune the first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value to obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and combine the first parameter value and the second parameter value corresponding to each parameter to be adjusted to obtain each second parameter value set other than the target first parameter value set;

[0192] Based on the pre-trained pass rate prediction model, each predicted pass rate corresponding to each input second parameter value set is determined, and according to each predicted pass rate and the target pass rate, the parameter value in the parameter value set corresponding to the highest pass rate is determined as the target parameter value of the image acquisition device.

[0193] Furthermore, the processor 801 is specifically configured to determine, based on each predicted qualified rate and the target qualified rate, a parameter value in a parameter value set corresponding to a highest qualified rate as a target parameter value of the image acquisition device, including:

[0194] If the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate, performing at least one fine adjustment based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine adjustment value, and determining the target parameter value of the image acquisition device based on whether the target predicted qualified rate determined after two adjacent fine adjustments is less than the target predicted qualified rate of the previous one;

[0195] If the target predicted qualified rate is not greater than the target qualified rate, the first parameter value in the target first parameter value set is determined as the target parameter value of the image acquisition device.

[0196] Furthermore, the processor 801 is specifically configured to determine the target parameter value of the image acquisition device based on whether the target prediction pass rate determined after two adjacent fine-tunings is less than the target prediction pass rate of the previous one, including:

[0197] Based on the pass rate prediction model, the second parameter value after the last fine-tuning is predicted. If the predicted target prediction pass rate for the next time is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the target prediction pass rate for the next time, and the fine-tuning process and the prediction process are performed again until the predicted target prediction pass rate for the next time is less than the predicted target prediction pass rate for the previous time, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

[0198] Furthermore, the processor 801 is further configured to, after obtaining each second parameter value set by combining each parameter value corresponding to each parameter to be adjusted, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model, the method further includes:

[0199] Each second parameter value in each second parameter value set is normalized.

[0200] The communication bus mentioned in the electronic device mentioned above may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in the figure, but this does not mean that there is only one bus or only one type of bus.

[0201] The communication interface 802 is used for communication between the electronic device and other devices.

[0202] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk memory. Alternatively, the memory may be at least one storage device located away from the processor.

[0203] The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; it can also be a digital signal processing processor (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0204] Example 14:

[0205] Based on the above embodiments, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program executable by a processor. When the program is executed on the processor, the processor implements the following steps:

[0206] According to a sample set consisting of a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and a qualified rate, obtaining any first parameter value set and a corresponding qualified rate in the sample set;

[0207] Inputting each first parameter value in the first parameter value set into the original pass rate prediction model, and obtaining the output predicted pass rate corresponding to the first parameter value set;

[0208] According to the pass rate and predicted pass rate corresponding to the first parameter value set, the parameter values ​​of each parameter of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

[0209] Furthermore, the process of determining the corresponding relationship includes:

[0210] For each preset first parameter value set of each parameter to be adjusted, sending each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value;

[0211] Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number;

[0212] According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

[0213] Furthermore, after obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model, the method further includes:

[0214] Normalize each first parameter value in the first parameter value set.

[0215] Example 15:

[0216] Based on the above embodiments, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program executable by a processor. When the program is executed on the processor, the processor implements the following steps:

[0217] Determining a target first parameter value set corresponding to a target pass rate with the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate;

[0218] Fine-tune the first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value to obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and combine the first parameter value and the second parameter value corresponding to each parameter to be adjusted to obtain each second parameter value set other than the target first parameter value set;

[0219] Based on the pre-trained pass rate prediction model, each predicted pass rate corresponding to each input second parameter value set is determined, and according to each predicted pass rate and the target pass rate, the parameter value in the parameter value set corresponding to the highest pass rate is determined as the target parameter value of the image acquisition device.

[0220] Furthermore, determining, based on each predicted qualified rate and the target qualified rate, a parameter value in a parameter value set corresponding to a highest qualified rate as a target parameter value of the image acquisition device includes:

[0221] If the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate, performing at least one fine adjustment based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine adjustment value, and determining the target parameter value of the image acquisition device based on whether the target predicted qualified rate determined after two adjacent fine adjustments is less than the target predicted qualified rate of the previous one;

[0222] If the target predicted qualified rate is not greater than the target qualified rate, the first parameter value in the target first parameter value set is determined as the target parameter value of the image acquisition device.

[0223] Furthermore, determining the target parameter value of the image acquisition device based on whether the target prediction pass rate determined after two adjacent fine-tunings is less than the target prediction pass rate of the previous one includes:

[0224] Based on the pass rate prediction model, the second parameter value after the last fine-tuning is predicted. If the predicted target prediction pass rate for the next time is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the target prediction pass rate for the next time, and the fine-tuning process and the prediction process are performed again until the predicted target prediction pass rate for the next time is less than the predicted target prediction pass rate for the previous time, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

[0225] Furthermore, after obtaining each second parameter value set by combining each parameter value corresponding to each parameter to be adjusted, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model, the method further includes:

[0226] Each second parameter value in each second parameter value set is normalized.

[0227] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0228] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0229] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0230] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0231] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for determining parameter values ​​of an image acquisition device, characterized in that: The method comprises: Determining a target first parameter value set corresponding to a target pass rate with the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate; Fine-tune the first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value to obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and combine the first parameter value and the second parameter value corresponding to each parameter to be adjusted to obtain each second parameter value set other than the target first parameter value set; Based on a pre-trained pass rate prediction model, each predicted pass rate corresponding to each input second parameter value set is determined; if the highest target predicted pass rate among each predicted pass rate is greater than the target pass rate, at least one fine-tuning is performed based on the second parameter value in the target second parameter value set corresponding to the target predicted pass rate and the fine-tuning value; and according to whether the target predicted pass rate determined after two adjacent fine-tunings is less than the previous target predicted pass rate, the target parameter value of the image acquisition device is determined; If the target predicted qualified rate is not greater than the target qualified rate, the first parameter value in the target first parameter value set is determined as the target parameter value of the image acquisition device.

2. The method according to claim 1, characterized in that Determining the target parameter value of the image acquisition device based on whether the target prediction pass rate determined after two adjacent fine-tunings is less than the target prediction pass rate of the previous one includes: Based on the pass rate prediction model, the second parameter value after the last fine-tuning is predicted. If the predicted target prediction pass rate for the next time is not less than the target prediction pass rate obtained in the previous prediction, the target second parameter value set is updated according to the second parameter value set corresponding to the target prediction pass rate for the next time, and the fine-tuning process and the prediction process are performed again until the predicted target prediction pass rate for the next time is less than the predicted target prediction pass rate for the previous time, and the second parameter value in the second parameter value set corresponding to the target prediction pass rate obtained in the previous prediction is determined as the target parameter value of the image acquisition device.

3. The method according to claim 1, characterized in that After obtaining each second parameter value set by combining each parameter value corresponding to each parameter to be adjusted, and before determining each predicted pass rate corresponding to each input second parameter value set based on the pre-trained pass rate prediction model, the method further includes: Each second parameter value in each second parameter value set is normalized.

4. The method according to claim 1, wherein The training process of the pass rate prediction model includes: According to a sample set consisting of a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and a qualified rate, obtaining any first parameter value set and a corresponding qualified rate in the sample set; Inputting each first parameter value in the first parameter value set into the original pass rate prediction model, and obtaining the output predicted pass rate corresponding to the first parameter value set; According to the qualified rate and the predicted qualified rate corresponding to the first parameter value set, the parameter values ​​of the parameters of the original qualified rate prediction model are adjusted to obtain a trained qualified rate prediction model.

5. The method according to claim 4, characterized in that The process of determining the corresponding relationship includes: For each preset first parameter value set of each parameter to be adjusted, sending each first parameter value in the first parameter value set to the image acquisition device, so that the parameter value of each parameter to be adjusted in the image acquisition device is the corresponding first parameter value; Obtaining a first number of first target images of a target object that meet a preset image condition and are captured by the image acquisition device, wherein the first target images are captured by the image acquisition device through real-time recognition of video data within a preset time period based on a preset algorithm; and determining a pass rate corresponding to the first parameter value set based on the first number and a pre-stored second number; According to the qualified rate corresponding to each first parameter value set, a corresponding relationship between each first parameter value set and the qualified rate is established.

6. The method according to claim 4, characterized in that After obtaining any first parameter value set and the corresponding pass rate in the sample set, and before inputting each first parameter value in the first parameter value set into the original pass rate prediction model, the method further includes: Normalize each first parameter value in the first parameter value set.

7. A device for determining parameter values ​​of an image acquisition device, characterized in that: The device comprises: a determination module, configured to determine a target first parameter value set corresponding to a target pass rate having the highest pass rate based on a predetermined correspondence between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate; a fine-tuning module, configured to fine-tune a first parameter value of each parameter to be adjusted in the target first parameter value set based on a preset fine-tuning value, obtain a set number of second parameter values ​​after fine-tuning on both sides of the first parameter value, and obtain each second parameter value set other than the target first parameter value set by combining the first parameter value and the second parameter value corresponding to each parameter to be adjusted; A prediction module, configured to determine each predicted pass rate corresponding to each input second parameter value set based on a pre-trained pass rate prediction model; The determining module is further configured to determine, based on each predicted qualified rate and the target qualified rate, a parameter value in a parameter value set corresponding to a highest qualified rate as a target parameter value of the image acquisition device; The determination module is specifically used to perform at least one fine-tuning based on the second parameter value in the target second parameter value set corresponding to the target predicted qualified rate and the fine-tuning value if the highest target predicted qualified rate among each predicted qualified rate is greater than the target qualified rate; and determine the target parameter value of the image acquisition device based on whether the target predicted qualified rate of the latter one is less than the target predicted qualified rate of the previous one among the target predicted qualified rates determined after two adjacent fine-tunings; if the target predicted qualified rate is not greater than the target qualified rate, determine the first parameter value in the target first parameter value set as the target parameter value of the image acquisition device.

8. The device according to claim 7, characterized in that The device comprises: an acquisition module, configured to acquire any first parameter value set and the corresponding pass rate in a sample set consisting of a predetermined correspondence relationship between each first parameter value set of each parameter to be adjusted of the image acquisition device and the pass rate; A training module is used to input each first parameter value in the first parameter value set into the original pass rate prediction model to obtain the predicted pass rate corresponding to the output first parameter value set; according to the pass rate and predicted pass rate corresponding to the first parameter value set, the parameter values ​​of each parameter of the original pass rate prediction model are adjusted to obtain a trained pass rate prediction model.

9. An electronic device, characterized in that: include: A processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; A computer program is stored in the memory. When the program is executed by the processor, the processor executes the computer program stored in the memory to implement the method for determining the parameter value of the image acquisition device as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that It stores a computer program, which, when executed by a processor, implements the method for determining the parameter value of an image acquisition device as claimed in any one of claims 1 to 6.

Images