A Depth Estimation Method Based on Continuous-Time Flow Fields

Abstract

The present invention relates to the technical field of image data processing, and particularly relates to a depth estimation method based on a continuous-time flow field. The method includes: acquiring traffic image information during the driving process of a vehicle; performing feature processing on the traffic image information to obtain a feature space; determining a feature change trend in a time series based on the feature spaces corresponding to two adjacent moments; performing depth prediction on a target object during the driving process based on the feature change trend to obtain depth feature information of the target object; and identifying the target object based on the depth feature information and the feature space at the current moment. This method combines the feature space with the displacements of the target object at different time points during the driving process, and can accurately calculate the depth feature information of the target object; it is convenient to identify the target object through the depth feature information and the feature space, so as to improve the accuracy of target object identification.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to a depth estimation method based on a continuous-time flow field in the technical field of image data processing. Background Art

[0002] At present, although the technology of trustworthy Artificial Intelligence (AI) develops rapidly, the implementation of trustworthy AI technology faces various challenges, including principle conflicts, a large amount of resource and time consumption, the complexity of working conditions (such as: the diversity of application scenarios and regulatory and legal differences, etc.).

[0003] In related technologies, the methods for vehicle depth estimation include: traditional depth estimation and depth estimation using neural networks. However, whether it is traditional depth estimation or depth estimation using neural networks, during the implementation process, the reliable target recognition conditions are relatively harsh, so it is difficult to ensure the accuracy of depth estimation. Summary of the Invention

[0004] The purpose of the present invention is to provide a depth estimation method based on a continuous-time flow field, and the specific technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present invention provides a depth estimation method based on a continuous-time flow field, and the method includes:

[0006] Obtain traffic image information of a vehicle during driving;

[0007] Perform feature processing on the traffic image information to obtain a feature space;

[0008] Based on the feature spaces corresponding to two adjacent moments, determine the feature change trend in the time series;

[0009] Based on the feature change trend, perform depth prediction on the target object during the driving process to obtain the depth feature information of the target object;

[0010] Based on the depth feature information and the feature space at the current moment, identify the target object.

[0011] In a second aspect, a depth estimation system based on a continuous-time flow field is provided, and the system includes:

[0012] An acquisition module, configured to obtain traffic image information of a vehicle during driving;

[0013] A processing module, configured to perform feature processing on the traffic image information to obtain a feature space;

[0014] A determination module, configured to determine a feature change trend in a time series based on feature spaces corresponding to two adjacent moments;

[0015] A prediction module, configured to perform in-depth prediction on a target object during the driving process based on the feature change trend to obtain depth feature information of the target object;

[0016] An identification module, configured to identify the target object based on the depth feature information and the feature space at the current moment.

[0017] In a third aspect, a computer program product is provided, which includes: computer program code that, when running on a computer, causes the computer to execute the method in the first aspect or any one of the possible implementation manners described in the first aspect.

[0018] In a fourth aspect, a computer-readable storage medium is provided, which stores computer program code that, when running on a computer, causes the computer to execute the method in the first aspect or any one of the possible implementation manners described in the first aspect.

[0019] The present invention has the following beneficial effects: After obtaining traffic image information of a vehicle during driving, feature processing is performed on the traffic image information to obtain a feature space, and based on the feature spaces corresponding to two adjacent moments, a feature change trend is determined in a time series; in this way, the change situation of the traffic image information in the time series can be accurately reflected through the feature change trend. Then, in-depth prediction is performed on the target object during the driving process based on the feature change trend to obtain depth feature information of the target object; in this way, the displacement of the target object at different time points during the driving process can be characterized by combining the feature change trend, so that the depth feature information of the target object can be accurately calculated; it is convenient to identify the target object through the depth feature information and the feature space at the current moment to improve the accuracy of target object identification. Description of the Drawings

[0020] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for describing the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a schematic diagram of the implementation principle of a vehicle depth estimation method in the related art;

[0022] Figure 2It is a schematic diagram of another implementation principle of the vehicle depth estimation method in the related art;

[0023] Figure 3 It is a schematic diagram of yet another implementation principle of the vehicle depth estimation method in the related art;

[0024] Figure 4 It is a schematic diagram of the implementation process of a depth estimation method based on a continuous time flow field provided by an embodiment of the present invention;

[0025] Figure 5 It is a schematic diagram of the implementation principle of a depth estimation method based on a continuous time flow field provided by an embodiment of the present invention;

[0026] Figure 6 It is a schematic diagram of yet another implementation principle of a depth estimation method based on a continuous time flow field provided by an embodiment of the present invention;

[0027] Figure 7 It is a schematic diagram of another implementation principle of a depth estimation method based on a continuous time flow field provided by an embodiment of the present invention;

[0028] Figure 8 It is a schematic diagram of the application scenario of a depth estimation method based on a continuous time flow field provided by an embodiment of the present invention;

[0029] Figure 9 It is a schematic diagram of the composition structure of a depth estimation system based on a continuous time flow field provided by an embodiment of the present invention;

[0030] Figure 10 It is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. Detailed implementation manners

[0031] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, details the specific implementation manners, structures, features and effects of a depth estimation method based on a continuous time flow field proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Among them, in the description of the embodiments of the present invention, unless otherwise specified, " / " means "or". For example, A / B can represent A or B. The "and / or" in the text is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of the present invention, "a plurality of" means two or more than two.

[0033] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs.

[0035] Trusted AI technology is developing rapidly. However, there are numerous difficulties on the way to practical application. On the one hand, it is not easy to achieve multiple principles such as robustness and interpretability. Conflicts frequently occur among various principles. Taking deep learning as an example, adversarial training aimed at improving robustness is very likely to come at the cost of fairness. At the same time, implementing trusted AI requires a huge amount of computing resources and a long time investment. On the one hand, reliable AI is like a big tree relying on high-quality soil and has an extreme thirst for high-quality data. Once the data has quality problems, filled with poor or biased information, the performance and accuracy of the model will be greatly reduced, with numerous loopholes in actual use and even possibly misleading results. Therefore, deep learning requires data collection, cleaning, annotation, and subsequent rigorous testing and verification, all of which are time-consuming and laborious. On the other hand, as the model scale and complexity of deep learning continue to climb, the computing resources required for training and inference also increase significantly, bringing an unimaginable cost pressure. Its high cost can discourage many enterprises. Moreover, the complexity of working conditions is another difficult obstacle. The working conditions are complex and changeable, including factors such as the working state of equipment, environmental conditions, load changes, combinations of AI with different intelligence levels and various applications, running time, and interactions with other systems. It is simply impossible to collect all working conditions, which makes the interpretability of trusted AI even more difficult.

[0036] In related technologies, for the depth estimation of a vehicle, it can be achieved by Figure 1 the depth estimation method shown as follows: First, obtain the camera installation height H (camera installation Height), focal length f, and the height h of the target object in the image; where: That is In this depth estimation method, it is necessary to ensure that the target is on the ground, the ground is flat, and the vehicle moves stably without pitching motion. In this way, the depth estimation method has the following limitations: 1. The limitation that the target is on the ground. Ground condition limitations: Conditions such as the material, hardness, and humidity of different grounds may affect the accuracy of detection results. For example, when detecting on soft muddy ground, the sinking of the wheels and the deformation of the ground may not be able to ensure that the target is on the ground; 2. The limitation of ground flatness, it is difficult to ensure absolute flatness: In actual detection, it is very difficult to find a completely flat ground without slope. Even if the ground is leveled, there may be slight slopes or uneven areas, which will have a certain impact; 3. The limitation that the vehicle moves stably without pitching motion, it is difficult to completely eliminate the pitching motion: In actual detection, due to the influence of the vehicle's own weight distribution, the elasticity of the suspension system, and external excitations (such as wind resistance, road unevenness, etc.), it is very difficult to completely eliminate the pitching motion of the vehicle.

[0037] In related technologies, for the depth estimation of vehicles, it can be achieved by Figure 2 the depth estimation of the neural network shown: Depth distribution refers to estimating the depth information of each image feature point in three-dimensional space to obtain the probability distribution of these points at different depths. The LSS ((Lift, Splat, Shoot)) algorithm, as a deep learning model, especially using Convolutional Neural Networks (CNN), etc., depends on large-scale datasets to learn the mapping relationship from the input image to the output depth map or other relevant information (such as Figure 2 the mapping relationship architecture described). Therefore, in order to train a high-performance LSS model, it is usually necessary to collect a large amount of labeled data and ensure that these data have sufficient diversity and accuracy. And this is also the limitation of LSS; a large amount of labeled data is required to obtain the depth distribution of the real world, and its reliability depends on the labeled data seen.

[0038] The basic idea of human depth estimation, as Figure 3 shown: That is where B represents the baseline distance, that is, the horizontal distance between the two eyes. D: Depth represents the vertical distance from the target object to the camera. F represents the focal length; Δy represents the difference between y c and y e ; y c represents the distance between the intersection of line 31 and line 32, and the intersection of line 33 and line 32; represents the distance between the intersection of line 34 and line 32, and the intersection of line 35 and line 32.

[0039] In the embodiments of the present invention, different displacement times are used to predict depth. In a moving scenario, such as in autonomous vehicle or robot navigation, the depth information of the target object will change dynamically as the observation position changes. By recording and analyzing the displacements of the target object at different time points, we can use this displacement information to predict its depth. This method can more accurately estimate the depth of the target object by comparing the images or depth data at different time points, thereby improving the resolvable target recognition of the vehicle.

[0040] The following specifically describes the specific solution of a depth estimation method based on a continuous time flow field provided by the present invention with reference to the accompanying drawings. Please refer to Figure 4 , which shows a schematic diagram of the implementation process of a depth estimation method based on a continuous time flow field provided by an embodiment of the present invention. The method includes:

[0041] 401. Obtain traffic image information during the driving of the vehicle.

[0042] Here, the traffic image information can be a continuous video stream collected by the vehicle-mounted camera of the vehicle during the driving of the vehicle.

[0043] 402. Perform feature processing on the traffic image information to obtain a feature space.

[0044] Here, the feature space refers to the vector space composed of all possible feature vectors. By using a convolutional residual network (CNN+ResNet) to perform two-dimensional (2D) feature extraction on the traffic image information and performing feature fusion on the extracted features, a feature space is obtained.

[0045] In some possible implementation manners, first, traffic image features are obtained by performing feature extraction on the traffic image information; for example, the edges and contours of the traffic image information are extracted by the 2D feature extraction network in the CNN+ResNet network to obtain traffic image features. The 2D feature extraction network is pre-trained based on the traditional backpropagation method. The CNN+ResNet network uses a U-Net architecture to reproduce and enhance the original information. To reduce the amount of calculation and training, the CNN+ResNet has fewer layers and shares parameter information when necessary compared with a general network.

[0046] Then, feature fusion is performed on the traffic image features to obtain the feature space. Here, the 2D feature information network performs feature fusion, and features at different levels are respectively matched during feature fusion, and features at the same level are optimized and combined. For example, features at different levels are fused through top-down and lateral connection methods, and then prediction is performed. For lateral connection: at each layer of the top-down path, operations such as adding or concatenating the upsampled feature map with the feature map at the corresponding level in the bottom-up path are performed to fuse feature information at different levels and obtain a feature map with higher quality. In this way, feature fusion can improve the performance of the model, adapt to complex scenarios, enhance the generalization ability of the model, etc. Feature fusion refers to the optimization and combination of different feature vectors extracted from the same pattern to form a more descriptive and discriminative feature set. These features can come from different sensors, different time periods, different processing levels, or different data types. By fusing these features, richer information can be captured, thereby improving the performance, accuracy, and robustness of the model, and further improving the performance of the model, adapting to complex scenarios, enhancing the generalization ability of the model, and so on.

[0047] 403, Based on the feature spaces corresponding to two adjacent moments, determine the feature change trend in the time series.

[0048] Here, after performing feature fusion on the traffic image features, first perform feature regression on the obtained feature space, and then based on the spatial features after feature regression at two adjacent moments, determine the feature change trend in the time series. The 2D feature information network performs feature regression, extracts features that have an impact on the target variable, and performs standardization or normalization processing on the features to ensure that they are on the same scale. The purpose of feature regression is to predict the value of the target variable based on the existing feature data.

[0049] Such as Figure 5 As shown, for the input image space 51 (i.e., traffic image information), feature extraction is performed, then feature fusion is performed to obtain the feature space; after that, feature regression is performed on the obtained feature space to respectively obtain the feature space 52 at time t and the feature space 53 at time (t - 1) (i.e., the feature spaces at two adjacent moments). By performing feature difference on the feature spaces 52 and 53 at two adjacent moments, the feature change trend in the time series can be obtained.

[0050] In some possible implementation manners, first, perform feature difference on the feature spaces corresponding to every two adjacent moments to obtain a plurality of difference features; and based on the plurality of difference features, determine the feature change trend in the time series. Such as Figure 5As shown, by performing feature difference on the feature space 52 and the feature space 53, multiple differential features can be obtained. By splicing the multiple differential features, the differential space 54 can be obtained. Through the feature space 54, the change trend of the feature information over time can be characterized, that is, the change trend of the feature in the time series is obtained. Here, a differential network is constructed based on the time flow, and differential operations are performed on the feature spaces at every two adjacent moments. After all the feature spaces are differentiated, multiple differential features are obtained, and the multiple differential features together constitute the differential network. The purpose of the differential operation is to capture the change trend or pattern of the feature information over time, so as to enhance the information.

[0051] 404, based on the feature change trend, perform depth prediction on the target object during the driving process to obtain the depth feature information of the target object.

[0052] Here, by analyzing the feature change trend in the time series, and then combining the imaging height difference of the target object in the image at adjacent moments, and the displacement of the vehicle between two adjacent moments, the depth feature information of the target object can be accurately calculated.

[0053] In some possible implementation manners, the above step 404 can be implemented through the following steps 441 to 443 (not shown in the figure):

[0054] 441, perform time feature analysis on the feature change trend to obtain the feature change trends at different moments.

[0055] Here, in the time series, by extracting the feature moments of the feature change trend, the feature change trends at different moments can be obtained. Since the feature change trend is characterized by the differential space, the feature change trends at different moments can be characterized by the differential features at different moments. In time series analysis, version difference (shift) is used to predict depth. In time series analysis, shift refers to moving the time series data forward or backward by a certain number of time steps. This operation can be used to generate lag variables, so as to analyze the autocorrelation of the time series data. Shift can be used to construct features, which can then be used to train a model to predict future depth values.

[0056] 442, based on the feature change trend at the current moment and the feature change trend at the previous moment, respectively determine the imaging height difference between the target object between the current moment and the previous moment, and the moving distance of the vehicle between the current moment and the previous moment.

[0057] In some possible implementation manners, by analyzing the difference between the first imaging height and the second imaging height in the traffic image information corresponding to the target object at adjacent moments, it is the imaging height difference; thus, the above step 442 can be implemented through the following process:

[0058] First, in the characteristic change trend at the previous moment, determine the first imaging height of the target object in the traffic image information corresponding to the previous moment.

[0059] Here, since the characteristic change trend is obtained through multiple differential features, the imaging height at each moment in the traffic image information can be obtained in the characteristic change trend at each moment. As Figure 6 shown, in the characteristic change trend at the previous moment (t + 1), obtain the height of the target object in the image, that is, the imaging height of the object on the imaging plane of the camera at this moment, that is, obtain the first imaging height h t+1 . In Figure 6 , H represents the target height of the target object, that is, the height of the object in the real scene. f represents the focal length, that is, the distance from the center of the camera lens to the imaging plane. D represents the distance from the object to the camera, that is, the depth distance. W represents the width of the object in the real scene (the width is used as a schematic of the scene dimension). s represents the moving distance between vehicle moments t and t + 1.

[0060] Secondly, in the characteristic change trend at the current moment, determine the second imaging height of the target object in the traffic image information corresponding to the current moment.

[0061] As Figure 7 shown, in the characteristic change trend at the current moment (t), obtain the height of the target object in the image, that is, obtain the second imaging height h t . Among them, H represents the target height of the target object, that is, the height of the object in the real scene. f represents the focal length, that is, the distance from the center of the camera lens to the imaging plane. D represents the distance from the object to the camera, that is, the depth distance. W represents the width of the object in the real scene (the width is used as a schematic of the scene dimension).

[0062] Finally, determine the imaging height difference between the first imaging height and the second imaging height.

[0063] In this way, by subtracting the first imaging height from the second imaging height, the imaging height difference between adjacent moments can be accurately obtained, so as to calculate the depth feature information by combining the imaging height difference with the moving distance between adjacent moments.

[0064] 443, based on the imaging height difference and the moving distance, obtain the depth feature information of the target object.

[0065] Here, by dividing the imaging height difference by the imaging height at the previous moment and then combining the moving distance, the depth feature information of the target object can be calculated. As Figure 5As shown, a deep network 55 is constructed through multiple differential features, and then, depth feature information of the target object is output through the deep network.

[0066] In some possible implementation manners, the above step 443 can be implemented through the following process:

[0067] First, obtain the first imaging height in the traffic image information corresponding to the target object at the previous moment.

[0068] Here, at the previous moment (i.e., the moment t + 1), the imaging height of the target object in the traffic image information, that is, the first imaging height.

[0069] Secondly, determine the ratio between the first imaging height and the imaging height difference.

[0070] Here, calculate the quotient between the first imaging height and the imaging height difference, that is, obtain the ratio.

[0071] Finally, fuse the ratio and the moving distance to obtain the depth feature information of the target object at the current moment.

[0072] Here, multiply the ratio and the moving distance to obtain the depth feature information of the target object at the current moment.

[0073] Here, first obtain the true height of the target object and the focal length corresponding to the current moment; then, based on the true height of the target object and the focal length corresponding to the current moment, fuse the ratio and the moving distance to obtain the depth feature information of the target object at the current moment. As Figure 6 and 7 shown, obtain the true height H and the focal length f of the target object, and the depth feature information D can be calculated through the following formulas (1) and (2):

[0074]

[0075] Based on the above formulas (1) and (2), it can be obtained that so as to obtain

[0076]

[0077] 405. Based on the depth feature information and the feature space at the current moment, identify the target object.

[0078] Here, construct multiple target recognition prediction heads through the depth feature information and combine with the feature space at the current moment, so as to be able to realize the accurate recognition of the target object.

[0079] In some possible implementation manners, step 405 above may be implemented through the following steps 451 to 453 (not shown in the figure):

[0080] 451. Based on the depth feature information, construct a target recognition prediction head.

[0081] For example, based on the depth feature information, construct a target recognition prediction head for realizing target recognition, obstacle recognition, motion feature prediction, navigable space recognition, semantic information, lane line recognition, traffic sign recognition, etc.

[0082] 452. Transmit the feature space at the current moment to the target recognition prediction head.

[0083] Here, feedback the feature space at the current moment to the target recognition prediction head, so that the target recognition prediction head can more accurately recognize the target object by combining the feature space at the current moment with the depth feature information.

[0084] 453. Use the target recognition prediction head to recognize the target object.

[0085] Here, after forming the target recognition prediction head through the depth feature information, the target recognition prediction head can accurately recognize the target object in the traffic image information.

[0086] In the embodiments of the present invention, after obtaining the traffic image information of the vehicle during driving, perform feature processing on the traffic image information to obtain a feature space, and based on the feature spaces corresponding to two adjacent moments, determine the feature change trend in the time series; in this way, the change situation of the traffic image information in the time series can be accurately reflected through the feature change trend. Then, based on the feature change trend, perform depth prediction on the target object during the driving process to obtain the depth feature information of the target object; in this way, combining the feature change trend can characterize the displacement of the target object at different time points during the driving process, so as to accurately calculate the depth feature information of the target object; it is convenient to recognize the target object through the depth feature information and the feature space at the current moment, so as to improve the accuracy of target object recognition.

[0087] In some possible implementation manners, the embodiments of the present invention provide a depth estimation method based on a continuous time flow field, which can be implemented through Figure 8 the following steps: First, obtain image data 801; second, perform depth estimation 802 using a model; then calculate the depth information of each object in the scene 803; and perform multi-target recognition 804, for example, obstacle recognition, lane line recognition, traffic sign recognition, etc.; finally, adjust the vehicle driving policy 805 in real time (such as, obstacle avoidance, lane change, parking, etc.).

[0088] An embodiment of the present invention provides a depth estimation system based on a continuous-time flow field. Refer to Figure 9 , which shows a schematic structural diagram of a depth estimation system based on a continuous-time flow field provided by an embodiment of the present invention. The system 900 includes:

[0089] An acquisition module 901, configured to acquire traffic image information during the driving of a vehicle;

[0090] A processing module 802, configured to perform feature processing on the traffic image information to obtain a feature space;

[0091] A determination module 903, configured to determine a feature change trend in a time series based on the feature spaces corresponding to two adjacent moments;

[0092] A prediction module 904, configured to perform depth prediction on a target object during the driving process based on the feature change trend to obtain depth feature information of the target object;

[0093] An identification module 905, configured to identify the target object based on the depth feature information and the feature space at the current moment.

[0094] In some possible implementation manners, the prediction module is further configured to perform time feature analysis on the feature change trend to obtain feature change trends at different moments; determine an imaging height difference between the target object between the current moment and the previous moment, and a moving distance of the vehicle between the current moment and the previous moment based on the feature change trend at the current moment and the feature change trend at the previous moment; and obtain the depth feature information of the target object based on the imaging height difference and the moving distance.

[0095] In some possible implementation manners, the prediction module is further configured to determine a first imaging height of the target object in the traffic image information corresponding to the previous moment in the feature change trend at the previous moment; and determine a second imaging height of the target object in the traffic image information corresponding to the current moment in the feature change trend at the current moment;

[0096] Determine an imaging height difference between the first imaging height and the second imaging height.

[0097] In some possible implementation manners, the prediction module is further configured to obtain a first imaging height of the target object in the traffic image information corresponding to the previous moment; determine a ratio between the first imaging height and the imaging height difference; and fuse the ratio and the moving distance to obtain the depth feature information of the target object at the current moment.

[0098] In some possible implementation manners, the prediction module is further configured to obtain the true height of the target object and the focal length corresponding to the current moment; based on the true height of the target object and the focal length corresponding to the current moment, fuse the ratio and the moving distance to obtain the depth feature information of the target object at the current moment.

[0099] In some possible implementation manners, the determination module is further configured to perform feature difference on the feature spaces corresponding to every two adjacent moments to obtain a plurality of difference features; based on the plurality of difference features, determine the feature change trend in the time series.

[0100] In some possible implementation manners, the recognition module is further configured to construct a target recognition prediction head based on the depth feature information; transmit the feature space at the current moment to the target recognition prediction head; and use the target recognition prediction head to recognize the target object.

[0101] In some possible implementation manners, the processing module is further configured to extract features from the traffic image information to obtain traffic image features; and perform feature fusion on the traffic image features to obtain the feature space.

[0102] Optionally, the transmission medium may be a wired link (such as, but not limited to, coaxial cable, optical fiber, and Digital Subscriber Line (DSL), etc.) or a wireless link (such as, but not limited to, Wireless Fidelity (WIFI), Bluetooth, and mobile device network, etc.). It should be noted that: for the system provided in the above embodiments, only the division of the above functional modules is used for illustration. In actual applications, the above functions may be allocated to different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be elaborated here.

[0103] Figure 10 is a schematic structural diagram of a computer device provided by an embodiment of the present invention. Exemplarily, as Figure 10 shown, the computer device 1000 includes: a memory 1001, a processor 1002, and a computer program 1003 stored in the memory 1001 and running on the processor 1002. When the processor 1002 executes the computer program 1003, the computer device can execute any one of the depth estimation methods based on continuous time flow fields introduced above.

[0104] In addition, an embodiment of the present invention also protects a system, which may include a memory and a processor. Among them, executable program code is stored in the memory, and the processor is configured to call and execute the executable program code to execute a depth estimation method based on a continuous-time flow field provided by an embodiment of the present invention. In this embodiment, the system can be divided into functional modules according to the above method examples. For example, it can correspond to each functional module, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is illustrative, only a logical function division, and there may be other division methods in actual implementation. It should be noted that all relevant contents of each step involved in the above method embodiment can be cited in the function description of the corresponding functional module, and will not be elaborated here.

[0105] It should be understood that the system provided in this embodiment is used to execute the above-mentioned depth estimation method based on a continuous-time flow field, so the same effect as the above implementation method can be achieved. In the case of adopting an integrated unit, the system may include a processing module and a storage module. Among them, when the system is applied to a device, the processing module can be used to control and manage the actions of the device. The storage module can be used to support the device to execute mutual program codes, etc. Among them, the processing module can be a processor or a controller, which can implement or execute various exemplary logical blocks, modules, and circuits described in combination with the disclosure of the present invention. The processor can also be a combination that realizes computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processing (DSP) and a microprocessor, etc. The storage module can be a memory.

[0106] In addition, the system provided by the embodiment of the present invention may specifically be a chip, a component, or a module. The chip may include a connected processor and a memory. Among them, the memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute a depth estimation method based on a continuous-time flow field provided by the above embodiment. This embodiment also provides a computer-readable storage medium, in which computer program code is stored. When the computer program code runs on a computer, the computer is enabled to execute the above-related method steps to implement a depth estimation method based on a continuous-time flow field provided by the above embodiment.

[0107] This embodiment also provides a computer program product. When the computer program product runs on a computer, it causes the computer to execute the above-related steps to implement a depth estimation method based on a continuous time flow field provided by the above embodiment. Among them, the system, computer-readable storage medium, computer program product, or chip provided in this embodiment are all used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, which will not be elaborated here. Through the description of the above embodiments, those skilled in the art can understand that for the convenience and conciseness of description, only the above division of each functional module is used as an example. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the system is divided into different functional modules to complete all or part of the functions described above. In the embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the division of modules or units is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the system or unit can be in an electrical, mechanical or other form.

[0108] It should be noted that: the above sequence of embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous. Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other. Each embodiment focuses on the differences from other embodiments. The above content is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered by the protection scope of the present invention.

Claims

1. A depth estimation method based on a continuous-time flow field, characterized in that, The depth estimation method based on continuous-time flow fields includes: Obtaining traffic image information during the vehicle's driving process; Performing feature processing on the traffic image information to obtain a feature space; Based on the feature spaces corresponding to two adjacent moments, determining the feature change trend in the time series; Performing time feature analysis on the feature change trend to obtain the feature change trends at different moments; Based on the feature change trend at the current moment and the feature change trend at the previous moment, respectively determining the imaging height difference of the target object during the driving process between the current moment and the previous moment, and the moving distance of the vehicle between the current moment and the previous moment; Based on the imaging height difference and the moving distance, obtaining the depth feature information of the target object; Based on the depth feature information, constructing a target recognition prediction head; Transmitting the feature space at the current moment to the target recognition prediction head; Using the target recognition prediction head to recognize the target object.

2. The depth estimation method based on a continuous-time flow field according to claim 1, characterized in that, The step of respectively determining the imaging height difference of the target object between the current moment and the previous moment, and the moving distance of the vehicle between the current moment and the previous moment based on the feature change trend at the current moment and the feature change trend at the previous moment includes: In the feature change trend at the previous moment, determining the first imaging height of the target object in the traffic image information corresponding to the previous moment; [[ID= ​ 3. The depth estimation method based on a continuous-time flow field according to claim 1, wherein, ​ ​ ​ ​ 4. A depth estimation method based on a continuous-time flow field according to claim 3, characterized in that ​ ​ ​ 5. A depth estimation method based on a continuous-time flow field according to claim 1, characterized in that, ​ ​ ​ 6. A depth estimation method based on a continuous-time flow field according to claim 1, characterized in that ​ ​ ​ 7. A depth estimation system based on a continuous-time flow field, characterized in that, ​ An acquisition module, configured to acquire traffic image information of a vehicle during driving; A processing module, configured to perform feature processing on the traffic image information to obtain a feature space; A determination module, configured to determine a feature change trend in a time series based on the feature spaces corresponding to two adjacent moments; A prediction module, configured to perform time feature analysis on the feature change trend to obtain feature change trends at different moments; based on the feature change trend at the current moment and the feature change trend at the previous moment, respectively determine an imaging height difference of a target object during the driving process between the current moment and the previous moment, and a moving distance of the vehicle between the current moment and the previous moment; based on the imaging height difference and the moving distance, obtain depth feature information of the target object; An identification module, configured to construct a target identification prediction head based on the depth feature information; transmit the feature space at the current moment to the target identification prediction head; and use the target identification prediction head to identify the target object.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program code, which when running on a computer causes the computer to execute the depth estimation method based on a continuous time flow field according to any one of claims 1 to 6.

Images