Drill pipe identification method and system, electronic device

Abstract

The application relates to a drill pipe identification method and system and an electronic device, the method comprising: acquiring original video data collected in a seismic field acquisition project; wherein the original video records a dynamic operation process of a drill pipe in a field operation scene, and contains real-time motion pictures of the drill pipe and interaction information between an operator and the drill pipe; pre-processing the original video; wherein the pre-processing comprises anti-shake processing; based on the pre-processed video data, analyzing an image change rate of the video data, determining a stable interval according to the image change rate; retaining video data corresponding to a time period belonging to the stable interval, and eliminating video data corresponding to a time period belonging to an interval outside the stable interval; and identifying and position tracking the drill pipe in the video data in the stable interval through a pre-trained deep learning model.

Description

Technical Field

[0001] This application relates to the field of video recognition technology for earthquake field acquisition projects, and in particular to a drill pipe recognition method and system, and electronic equipment. Background Technology

[0002] In seismic field acquisition projects, real-time video monitoring of field operations is often necessary. While video recognition technology is a key means to improve exploration efficiency and safety during the digital transformation of seismic field acquisition operations, its practical application still faces multiple technical challenges. First, the inconsistent quality of videos, especially poor video clarity and stability, directly weakens the reliability and effectiveness of video recognition data. Second, drill rods, as core operational equipment, dynamically change shape with the operational stage (e.g., upright, horizontal, semi-buried), and their movement trajectory exhibits non-linear characteristics due to mechanical operation, making it difficult for traditional image processing algorithms to accurately capture changes in their quantity and spatial position. Furthermore, existing technologies rely on manual verification of drill rod status, resulting in low efficiency, poor real-time performance, and high subjective error rates, failing to meet the quality control requirements of high-intensity exploration operations. However, customizing a drill rod identification solution requires customized development across the entire chain, from data annotation to model training, which is technically complex, time-consuming, and costly. Therefore, how to automatically and accurately identify drill rods and their positional changes has become an urgent problem to be solved. Summary of the Invention

[0003] This application provides a drill pipe identification method and system to solve the problem of how to accurately identify a specific object, such as a drill pipe, and its dynamic changes in a complex and ever-changing natural environment.

[0004] Firstly, this application provides a drill rod identification method, which includes: acquiring raw video data collected in a seismic field acquisition project; wherein the raw video records the dynamic operation process of the drill rod in a field operation scenario, including real-time motion images of the drill rod and interaction information between the operator and the drill rod; preprocessing the raw video; wherein the preprocessing includes image stabilization; based on the preprocessed video data, analyzing the image change rate of the video data, and determining a stable interval according to the image change rate; retaining the video data corresponding to the time segment to which the stable interval belongs, and removing the video data corresponding to the time segment outside the stable interval; and identifying and tracking the position of the drill rod in the video data within the stable interval using a pre-trained deep learning model.

[0005] As an optional implementation, the image stabilization process includes stabilizing the original video using ORB technology.

[0006] As an optional implementation, the step of analyzing the image change rate of the preprocessed video data and determining a stable interval based on the image change rate includes: dividing the video into multiple regions based on the preprocessed video data, statistically analyzing the image change rate of the multiple regions in chronological order, and generating change rate curves for the multiple regions; analyzing the change rate curves using an unsupervised machine learning algorithm; and classifying the curves according to their volatility, grouping those with low volatility as stable intervals.

[0007] As an optional implementation, the step of identifying and tracking the drill rod in video data within the stable interval using a pre-trained deep learning model includes: using a pre-trained deep learning model to outline the boundary of the drill rod in video data within the stable interval, and then identifying and tracking the drill rod.

[0008] As an optional implementation, the method further includes: identifying the posture of the operator in the video data.

[0009] As an optional implementation, the method further includes: extracting time-series data of the length and angle changes of the drill rod and time-series data of the operator's posture and movements; converting the time-domain signal corresponding to the time-series data into a frequency-domain signal through Fourier transform, and analyzing the periodic characteristics of the length, angle, and posture and movements of the drill rod; based on the periodic characteristics, associating the synchronicity of the length and angle changes of the drill rod with the posture and movements of the operator, and determining the number and position of the drill rod.

[0010] As an optional implementation, the deep learning model employs data augmentation techniques during the training phase, including random cropping, rotation, and color changes.

[0011] As an optional implementation, the method further includes performing brightness equalization processing on the original video, wherein the brightness equalization processing includes adjusting the brightness distribution of video frames in the original video to enhance the contrast between the drill rod and the background environment, so as to highlight the shape characteristics of the drill rod.

[0012] As an optional implementation, the method further includes identifying and removing useless segments from the original video.

[0013] Secondly, this application provides a drill rod identification system, comprising: a first module for acquiring raw video data collected in a seismic field acquisition project; wherein the raw video records the dynamic operation process of the drill rod in a field operation scenario, including real-time motion images of the drill rod and interaction information between the operator and the drill rod; a second module for preprocessing the raw video; wherein the preprocessing includes image stabilization; a third module for analyzing the image change rate of the preprocessed video data and determining a stable interval based on the image change rate; a fourth module for retaining the video data corresponding to the time segment to which the stable interval belongs and removing the video data corresponding to the time segment outside the stable interval; and a fifth module for identifying and tracking the drill rod's position in the video data within the stable interval using a pre-trained deep learning model.

[0014] Thirdly, the application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to implement the drill pipe identification method as described above when executing the program stored in the memory.

[0015] The technical solutions provided in this application have the following advantages compared with the prior art:

[0016] The drill rod identification method and system provided in this application eliminate video blur and noise caused by camera shake and uneven lighting through preprocessing such as image stabilization, providing clear input data for subsequent analysis. Through rate of change analysis and stable interval filtering, only high-quality video segments are retained, resulting in high-quality video data. This ensures video clarity and stability, thereby guaranteeing the reliability and effectiveness of subsequent analysis. The method provided in this application achieves accurate identification of specific objects such as drill rods, greatly improving the efficiency of field operations. Especially in complex and changeable natural environments, it can accurately identify drill rods and their dynamic changes, automatically and accurately count the number and location of drill rods, improve work efficiency, and reduce human error. This is of great significance for guiding work processes and preventing accidents. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0018] Figure 1 This illustration schematically shows the system architecture of the drill pipe identification method and system applicable to embodiments of this application;

[0019] Figure 2 This is a schematic flowchart of a drill pipe identification method according to an embodiment of this application;

[0020] Figure 3 This is a schematic diagram of image change rate analysis of a drill pipe identification method according to an embodiment of this application; Figure 3 It displays statistics on the rate of change in different regions of the video, used to identify the stable intervals of the video.

[0021] Figure 4 This is a schematic diagram of a drill pipe identification method according to an embodiment of this application;

[0022] Figure 5 The drill pipe angle variation curve is shown in the drill pipe identification method according to an embodiment of this application.

[0023] Figure 6 The drill pipe length variation curve is shown in another embodiment of the drill pipe identification method according to this application.

[0024] Figure 7 The drill pipe angle variation curve is shown in another embodiment of the drill pipe identification method according to this application.

[0025] Figure 8 The operator posture change curve is shown in another embodiment of the drill pipe identification method according to this application.

[0026] Figure 9 To extract curves according to the rules of the drill pipe identification method according to another embodiment of this application;

[0027] Figure 10 A model performance evaluation diagram of a drill pipe identification method according to another embodiment of this application;

[0028] Figure 11 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0030] Figure 1The schematic illustration shows the system architecture of the drill pipe identification method and system applicable to embodiments of this application.

[0031] Reference Figure 1 As shown, the system architecture 100 of the drill pipe identification method and system applicable to embodiments of this application includes: terminal devices 101, 102, and 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the terminal devices 101, 102, and 103 and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0032] Users can interact with server 105 via network 104 using terminal devices 101, 102, and 103 to receive or send messages, etc. Terminal devices 101, 102, and 103 can be equipped with audio / video capture devices, image / audio / video playback applications, etc. Other communication client applications can also be installed, such as web browser applications, search applications, image storage and processing applications, instant messaging tools, email clients, social media platform software (for example only), etc. Terminal devices 101, 102, and 103 can be various electronic devices with displays and supporting software capable of implementing drill pipe recognition methods. These electronic devices can further include drill pipe recognition systems, such as smartphones, tablets, laptops, desktop computers, etc. Server 105 can be a server providing various services, such as a backend management server that processes and provides service support for the video data received and stored by users using terminal devices 101, 102, and 103 (for example only). The backend management server can analyze and process the received video data and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0033] It should be noted that the drill pipe identification method provided in this application embodiment can generally be executed by server 105 or a terminal device with a certain computing power. Correspondingly, the drill pipe identification system provided in this application embodiment can generally be set in server 105 or the aforementioned terminal device with a certain computing power. The drill pipe identification method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105. Correspondingly, the drill pipe identification system provided in this application embodiment can also be set in a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105. It should be understood that... Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0034] Figure 2 This is a flowchart illustrating the drill pipe identification method according to an embodiment of this application; as shown Figure 2 As shown in the figure, this application provides a drill pipe identification method, which includes the following steps:

[0035] S101, acquire the raw video data collected in the earthquake field acquisition project; wherein, the raw video records the dynamic operation process of the drill rod in the field operation scene, including the real-time movement of the drill rod and the interaction information between the operator and the drill rod;

[0036] S102, preprocess the original video; wherein, the preprocessing includes image stabilization.

[0037] S103, Based on the preprocessed video data, analyze the image change rate of the video data, and determine a stable interval according to the image change rate;

[0038] S104, retain the video data corresponding to the time segment to which the stable interval belongs, and remove the video data corresponding to the time segment to which the stable interval belongs;

[0039] S105, the drill rod will be identified and its position tracked in the video data within the stable range using a pre-trained deep learning model.

[0040] In S101, in practical application scenarios, the raw video data can be acquired by camera equipment deployed at the work site. For example, raw video data can be acquired synchronously through multi-view fixed cameras and wearable mobile devices. Specifically, the actual acquisition requirements depend on the project requirements and are not limited here. For example, the frame rate can be no less than 30fps, and the resolution no less than 1080p. The higher these acquisition parameters are required, the clearer the capture of the drill pipe's shape and motion characteristics can be guaranteed.

[0041] In one possible implementation, real-time video monitoring of field operations during seismic field acquisition projects can be used for drill pipe counting. For example, during project construction, the interaction between operators and drill pipes can involve workers repeatedly inserting and removing drill pipes to complete well drilling. When well depth measurement is required, the drill pipe identification method according to the embodiments of this application can be used to identify the drill pipes, thereby determining the number of drill pipes, and then accurately calculating the well depth by multiplying the number of drill pipes by the length of each pipe. In this application scenario, the drill pipe identification method can automatically and accurately count the number and location of drill pipes, significantly improving technical accuracy and work efficiency compared to manually watching videos and counting the number of drill pipe insertions and removals.

[0042] In S102, in practical application scenarios, preprocessing may include brightness equalization processing. The video footage may include not only the drill rod and operators, but also outdoor scenes such as dynamic backgrounds (e.g., wind-blown vegetation) and static targets (e.g., the drill rod). Brightness equalization processing can be used to adjust the brightness distribution of video frames, enhancing the contrast between the drill rod and the background environment to highlight the drill rod's morphological features and provide optimized image data for subsequent recognition and tracking. For example, brightness equalization processing can be implemented by adjusting video brightness, enhancing contrast, and highlighting the rod's features. Considering the outdoor scene, lighting conditions are variable during operation; therefore, brightness equalization processing can also be implemented as global or local brightness correction to ensure that the drill rod features are clearly distinguishable in video frames under different lighting conditions, and that the light from each object in the video is visible. Brightness equalization processing can also be implemented at various stages of the drill rod recognition method according to the embodiments of this application.

[0043] For example, image stabilization can employ ORB (Object-Recognition Bypass) as a technical feature point recognition algorithm for video stabilization. ORB feature points are matched between consecutive frames, and by calculating the displacement vectors or motion trajectories of feature points between adjacent frames, the local or global motion (such as translation or rotation) of the video is indirectly reflected. Image stabilization can provide a data foundation for subsequent rate of change statistics.

[0044] In one possible implementation, the method further includes identifying and removing useless segments from the original video. For example, useless segments could be empty shots without any operation or violently shaking segments. In practical applications, various other methods can also be used for differentiation. A preliminary analysis based on the overall video change rate can be conducted to automatically identify and extract useful segments from the video, removing invalid segments. For instance, based on the pre-processed complete video (which has been stabilized and / or brightness equalized), the global change rate can be calculated frame by frame. Useless segments can be identified based on this change rate. For example, high change rate segments can be defined; if the global change rate exceeds a threshold (e.g., 20 pixels) for N consecutive frames (e.g., 10 frames), it is determined to be an invalid segment. Typical scenarios in practical applications include violent camera shaking, rapid movement of operators obstructing the lens, and sudden environmental interference (e.g., animals entering the scene). Based on this operation, invalid segments can be initially screened, avoiding subsequent steps on invalid segments, reducing computational resource waste, and preventing invalid segments from interfering with drill pipe identification and statistical results, laying the foundation for subsequent stable interval analysis and drill pipe identification.

[0045] In S103, in practical application scenarios, stable or dynamically significant regions in the preprocessed video are identified by statistically analyzing the image change rates (e.g., feature point motion amplitude, background stability, etc.) of different regions. This analysis provides a basis for subsequent segment extraction. By first determining stable regions through image change rate analysis, and then retaining stable regions while eliminating unstable regions, the amount of data processed subsequently can be reduced, improving overall efficiency.

[0046] In one possible implementation, S103 may include: dividing the video into multiple regions based on the preprocessed video data, statistically analyzing the image change rates of the multiple regions in chronological order, and generating change rate curves for the multiple regions; analyzing the change rate curves using an unsupervised machine learning algorithm; and classifying the curves according to their volatility, grouping those with low volatility as stable intervals. The above operations can be understood as distinguishing stable intervals (low change rate) from unstable intervals (high change rate) using an unsupervised learning method. For example, each preprocessed video frame is divided into several regions. For each region, the rate of change of adjacent frames is calculated in chronological order (video frame sequence). Then, a time-rate of change curve is generated for each region. The horizontal axis represents time (e.g., frame number, timestamp, etc.), and the vertical axis represents the rate of change value of that region (e.g., average pixel displacement, etc.). Each curve can reflect the degree of fluctuation of the rate of change of the corresponding region. Therefore, based on the volatility, unsupervised classification can be performed. Cluster analysis is performed on the rate of change curves of all regions. The mean, variance, peak frequency, and other features of the rate of change are input to obtain different categories. For example, the curves are divided into groups with high volatility and groups with low volatility according to the volatility. Then, the region corresponding to the low volatility group is selected as the candidate stable interval.

[0047] In S104, the stable interval can be used to provide high-quality video input for subsequent drill rod identification and tracking. Within the stable interval, the motion patterns of the drill rod (such as angle and displacement) are more easily extracted accurately. In complex and ever-changing natural environments, if the video is unstable, even if the drill rod appears in the frame, identification may fail due to poor data quality. By removing data from unstable intervals and substandard video data, the high quality of video data can be guaranteed, invalid data can be avoided, low-quality data can be prevented from interfering with the model, and the reliability and effectiveness of subsequent analysis can be guaranteed.

[0048] S105 involves using a pre-trained deep learning model to identify and track the position of the drill rod in the video data within the stable interval. In one possible implementation, S105 may include: using the pre-trained deep learning model to outline the boundary of the drill rod in the video data within the stable interval, and then identifying and tracking the drill rod's position. Specifically, the performance of the deep learning model can be optimized by adjusting its structure or parameters.

[0049] Specifically, deep learning models can employ data augmentation techniques during the training phase, including random cropping, rotation, and color transformation. Random cropping, rotation, and color space transformation can improve the model's robustness to changes in lighting and viewing angle. In one possible implementation, the model can be divided into training, testing, and validation sets in a 70%, 20%, and 10% ratio, respectively, for training the deep learning model to identify drill rods in videos. For example, the deep learning model can be generated by training on a specific dataset, which includes at least 70% for training, 20% for testing, and 10% for validation samples. Depending on the specific application scenario, the dataset can contain samples of drill rod operation scenarios under different lighting conditions and multiple angles.

[0050] Based on the above operations, the drill rod identification method provided in this application eliminates video blur and noise caused by camera shake and uneven lighting through preprocessing such as image stabilization, providing clear input data for subsequent analysis. Through rate of change analysis and stable interval filtering, it retains only high-quality video segments, thereby obtaining high-quality video data, ensuring video clarity and stability, and thus ensuring the reliability and effectiveness of subsequent analysis. The method provided in this application achieves accurate identification of specific objects such as drill rods, greatly improving the efficiency of field operations. Especially in complex and changeable natural environments, it can accurately identify drill rods and their dynamic changes, automatically and accurately count the number and location of drill rods, improve work efficiency and reduce human error, which is of great significance for guiding work processes and preventing accidents.

[0051] Furthermore, the drill rod identification method provided in this application embodiment may also include identifying the operator's posture in video data. It may further include the following operations: extracting time-series data of drill rod length and angle changes and time-series data of operator posture movements; converting the time-domain signal corresponding to the time-series data into a frequency-domain signal using Fourier transform, and analyzing the periodic characteristics of the drill rod length, angle, and operator posture movements; based on the periodic characteristics, associating the synchronicity between the drill rod length and angle changes and the operator's posture movements to determine the number and position of drill rods. In field operations, two drill rods may be installed at the same location successively, and the two drill rods may appear alternately. However, it is impossible to distinguish whether the same drill rod is being repeatedly operated based on a single frame. Therefore, based on the identification and position tracking of drill rods using a pre-trained deep learning model, the spatiotemporal association between the operator's posture and the drill rod movements can be used to avoid the effects of occlusion, false detection, and multi-target confusion, thereby reducing human error and improving statistical accuracy. After collecting data on drill pipe length and angle changes, as well as operator posture, across all stable intervals, the dynamic patterns of these changes are extracted and correlated with the aforementioned patterns to comprehensively determine the final number and position of drill pipes in relation to the operator's posture. This cross-dimensional data integration enables accurate identification of drill pipe quantity and position changes. Integrating multi-dimensional, multi-time-point data improves the overall accuracy of drill pipe quantity and position determination.

[0052] Based on the same inventive concept, this application also provides a drill rod identification system, which includes: a first module for acquiring raw video data collected in a seismic field acquisition project; wherein the raw video records the dynamic operation process of the drill rod in the field operation scene, including real-time motion images of the drill rod and interaction information between the operator and the drill rod; a second module for preprocessing the raw video; wherein the preprocessing includes image stabilization; a third module for analyzing the image change rate of the video data based on the preprocessed video data, and determining a stable interval based on the image change rate; a fourth module for retaining the video data corresponding to the time segment to which the stable interval belongs, and removing the video data corresponding to the time segment outside the stable interval; and a fifth module for identifying and tracking the position of the drill rod in the video data within the stable interval using a pre-trained deep learning model.

[0053] Based on the above embodiments, the drill pipe recognition system provided in this application eliminates video blur and noise caused by camera shake and uneven lighting through preprocessing such as image stabilization, providing clear input data for subsequent analysis. Through rate of change analysis and stable interval filtering, it retains only high-quality video segments, thus obtaining high-quality video data, ensuring video clarity and stability, and consequently guaranteeing the reliability and effectiveness of subsequent analysis. The system provided in this application achieves accurate recognition of specific objects such as drill pipes, greatly improving the efficiency of field operations. Especially in complex and changing natural environments, it can accurately identify drill pipes and their dynamic changes, automatically and accurately count the number and location of drill pipes, improve work efficiency, and reduce human error. This is of great significance for guiding work processes and preventing accidents. More details and beneficial effects of this embodiment can be found in the relevant descriptions of the foregoing embodiments, and will not be repeated here.

[0054] The drill pipe identification method of this application is illustrated below with some examples:

[0055] Practical Example 1

[0056] 1) Video preprocessing: Use the ORB algorithm for feature point recognition and perform image stabilization on the video; adjust the video brightness, enhance contrast, and highlight the pole features;

[0057] 2) Extract useful segments: Automatically identify and remove useless segments through video change rate analysis;

[0058] 3) Rate of change analysis: The stable intervals in the video are identified by statistical analysis of the rate of change of the preprocessed video.

[0059] Figure 3 This is a schematic diagram of image change rate analysis of a drill pipe identification method according to an embodiment of this application; Figure 3 It displays statistics on the rate of change in different regions of the video, used to identify stable intervals in the video. For example... Figure 3 As shown, the rate of change of images in different regions of the video is statistically analyzed in chronological order to generate a time series rate of change curve.

[0060] Specifically, Figure 3 The horizontal axis represents the frame rate, and the vertical axis represents the rate of change. Curves r1, r1, r1, r1, and r1 represent the pixel change rates of regions one, two, three, four, and five, respectively. Regions one, two, three, four, and five represent the five regions of a vertically shot video, divided from top to bottom. Figure 3It can be seen that the movement patterns in regions two and three are more pronounced, which aligns with the fact that workers in actual scenarios are typically located in these regions in the captured video, i.e., in the center position. In other words, each preprocessed video frame is divided into five independent regions (regions one through five) from top to bottom. For each region, the rate of change of adjacent frames is calculated in chronological order (video frame sequence) (e.g., using ORB feature point matching to calculate the average pixel displacement of matching point pairs). Then, a time-rate-change curve is generated for each region; the horizontal axis represents time (e.g., frame number, timestamp), and the vertical axis represents the rate of change value of that region (e.g., average pixel displacement). Each curve reflects the degree of fluctuation in the rate of change of the corresponding region. For example, combining... Figure 3 As can be seen, the curve in region 3 tends to flatten out in the middle of the time period, but fluctuates sharply around frame 1222 shown in the figure. Therefore, based on volatility, unsupervised classification can be performed by clustering the rate of change curves of all regions. By inputting features such as the mean, variance, and peak frequency of the rate of change, different categories can be obtained. For example, the curves can be divided into high volatility, medium volatility, and low volatility groups based on volatility (or, for example, into groups with high volatility and groups with low volatility). Then, the regions corresponding to the low volatility group are selected as candidate stable intervals. This implementation method does not rely on a fixed threshold, but automatically classifies the stability level through data distribution, providing a basis for subsequent stable interval determination.

[0061] 4) Object recognition: The drill rod in the video is identified using a deep learning model; a training dataset is prepared and divided into training, test, and validation sets according to a ratio of 70%, 20%, and 10%; the deep learning model is trained to identify the drill rod in the video; in this embodiment, the model's Box Precision is 0.707, Recall is 0.8, mAP50 is 0.809, and mAP50-95 is 0.414.

[0062] 5) Drill rod position tracking: Enables real-time position tracking of the drill rod in the video;

[0063] 6) Pattern Extraction: Extract the movement patterns of the drill pipe and the operator, including changes in the length and angle of the drill pipe, as well as changes in the operator's posture;

[0064] 7) Statistical analysis: By statistically analyzing the changing patterns of drill rods in the video, the number of drill rods can be determined.

[0065] Figure 4 The diagram shows the trajectory of the drill pipe angle changing over time, indicating that... Figure 4 The data shows that the bounding box of the drill pipe is accurately identified, thus clearly determining the position of the drill pipe and enabling identification and position tracking. Figures 5 to 8The data displayed includes curves showing changes in drill pipe length and angle, as well as curves showing changes in operator posture. These curves, including time-series data on drill pipe length and angle changes and operator posture movements, illustrate how each feature changes over time and are considered time-domain data. Figure 5 The curve showing the change of drill pipe angle l-angle over time is displayed. Figure 6 The curve showing the change of the height h of the frame (the frame refers to the boundary enclosed by the drill pipe) over time reveals a strong regularity, consistent with the regular insertion and removal of the drill pipe in actual application scenarios. Figure 7 The curve showing the operator's height changing over time exhibits strong regularity, consistent with the actual scenario of a person regularly bending over to pull up a drill rod. Figure 8 The curve showing the change of drill pipe length l-length over time exhibits strong regularity, consistent with the actual application scenario where one drill pipe is pulled out and set aside, and then the next drill pipe is pulled out. Figures 4 to 9 The feature changes shown are merely examples. In real-world applications, various features related to the operator and drill rod can be statistically analyzed, including but not limited to: the center point position of the rod, the rod angle, the height and width of the rod frame, confidence level (confidence is usually highest when the rod is fully pulled out), human posture (e.g., left hand x, left hand y, right hand x, right hand y), height, left shoulder x, left shoulder y, right shoulder x, right shoulder y, etc. The peak red dots on the curves in the graph indicate drastic changes at that point in time.

[0066] Figure 9 The diagram illustrates pattern extraction curves, including the RSY curve of a person's right shoulder over time. The blue curve represents the original data, where the regularity is not obvious. The orange curve represents trend fitting, and the green curve represents the result after subtracting the trend fitting from the original data, showing a stronger regularity. The diagram also includes frequency analysis of the variation patterns using Fourier transform, converting them into frequency domain signals. As seen in the pattern extraction curves, Fourier transform frequency analysis of the variation patterns demonstrates how Fourier transform converts the time-domain signals corresponding to time-series data into frequency domain signals. This analysis examines the periodic characteristics of the drill rod length, angle, and operator posture. Based on these periodic characteristics, the synchronicity between changes in drill rod length and angle and operator posture is correlated to determine the number of drill rods. For example, analyzing the time synchronicity between operator posture changes (e.g., arm raising) and drill rod movements (e.g., installation completion) enhances the reliability of data for that time period if the posture change is highly correlated with the drill rod state, which can be understood as verifying the correlation between the operator and drill rod. Through Fourier transform, frequency analysis of the dynamic changes in drill rods and operators can be achieved, thus aiding in more accurate statistical correlation analysis.

[0067] In the above operations, the changes in the pole and the person in the video are statistically analyzed, including but not limited to changes in the pole's length and angle, as well as changes in the operator's posture. In combination with specific application scenarios, other physical quantities can also be collected and analyzed for correlation analysis according to actual engineering needs.

[0068] Figure 10 (a)-(d) illustrate the performance of the pre-trained model, including precision, recall, and F1 score at different confidence thresholds, including precision-confidence curves, precision-recall curves, recall-confidence curves, and F1 score-confidence curves. Typically, the model is used to identify drill pipes when the performance shown in the figures is achieved.

[0069] Based on the above embodiments, the drill rod identification method provided in this application eliminates video blur and noise caused by camera shake and uneven lighting through preprocessing such as image stabilization, providing clear input data for subsequent analysis. Through rate of change analysis and stable interval filtering, only high-quality video segments are retained, thereby obtaining high-quality video data, ensuring video clarity and stability, and thus ensuring the reliability and effectiveness of subsequent analysis. The method provided in this application achieves accurate identification of specific objects such as drill rods, greatly improving the efficiency of field operations. Especially in complex and changeable natural environments, it can accurately identify drill rods and their dynamic changes, automatically and accurately count the number and location of drill rods, improve work efficiency and reduce human error, which is of great significance for guiding work processes and preventing accidents.

[0070] Furthermore, the drill pipe identification method provided in this application, based on the identification and position tracking of drill pipes using a pre-trained deep learning model, also avoids the influence of occlusion, false detection, and multi-target confusion by spatiotemporally associating the operator's posture with the drill pipe's movements, thereby reducing human error and improving statistical accuracy. After collecting the changes in drill pipe length, angle, and operator posture within all stable intervals, the dynamic change patterns of drill pipe length, angle, and operator posture are extracted and correlated with these patterns to comprehensively determine the final number and position of drill pipes in relation to the operator's posture. This achieves cross-dimensional data integration and accurate identification of drill pipe quantity and position changes. By integrating multi-dimensional and multi-time-point data, the global accuracy of drill pipe quantity and position determination can be improved. This effectively improves the accuracy and efficiency of drill pipe statistics, reduces the workload of staff, and minimizes errors caused by human factors.

[0071] Practical Example 2

[0072] This application also provides a drill pipe recognition system for drill pipe video recognition in seismic field acquisition projects. The system includes the following components: a video preprocessor for preprocessing the original video, including but not limited to image stabilization and brightness equalization; a useful segment extractor for extracting useful segments from the video; a rate of change analyzer for analyzing the rate of change in the video and identifying stable intervals; and a drill pipe recognizer for identifying and tracking the position of drill pipes in the video using a deep learning model. Optionally, the system may further include: an unsupervised machine learning module for classifying rate of change curves to identify stable intervals in the video. Optionally, the system may further include: a data augmentation module for performing data augmentation during model training to improve the model's generalization ability. Optionally, the system may further include: a pattern extractor for statistically analyzing the changing patterns of the drill pipe and the operator, including but not limited to changes in the length and angle of the drill pipe and changes in the operator's posture.

[0073] Based on the above embodiments, the drill pipe recognition system provided in this application eliminates video blur and noise caused by camera shake and uneven lighting through preprocessing such as image stabilization, providing clear input data for subsequent analysis. Through rate of change analysis and stable interval filtering, it retains only high-quality video segments, thus obtaining high-quality video data, ensuring video clarity and stability, and consequently guaranteeing the reliability and effectiveness of subsequent analysis. The system provided in this application achieves accurate recognition of specific objects such as drill pipes, greatly improving the efficiency of field operations. Especially in complex and changing natural environments, it can accurately identify drill pipes and their dynamic changes, automatically and accurately count the number and location of drill pipes, improving work efficiency and reducing human error. This is of great significance for guiding work processes and preventing accidents. Furthermore, the drill pipe recognition system provided in this application, based on the identification and position tracking of drill pipes using a pre-trained deep learning model, also avoids the influence of occlusion, false detection, and multi-target confusion by spatiotemporally associating the operator's posture with the drill pipe's movements, reducing human error and improving statistical accuracy. After collecting data on drill rod length and angle changes, as well as operator posture, across all stable intervals, the dynamic patterns of these changes are extracted and correlated with the aforementioned patterns to comprehensively determine the final number and position of drill rods in relation to the operator's posture. This cross-dimensional data integration enables accurate identification of drill rod quantity and position changes. Integrating multi-dimensional, multi-time-point data improves the overall accuracy of drill rod quantity and position determination. This effectively enhances the accuracy and efficiency of drill rod statistics, reduces the workload of staff, and minimizes errors caused by human factors. Further details and beneficial effects of this embodiment can be found in the descriptions of the foregoing embodiments and will not be repeated here.

[0074] Any number of the functional modules included in the above-described device can be combined into one module, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules can be combined with at least part of the functionality of other modules and implemented in one module. At least one of the functional modules included in the above-described device can be at least partially implemented as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a System-on-Chip, a System-on-Substrate, a System-on-Package, an Application-Specific Integrated Circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the functional modules included in the device can be at least partially implemented as a computer program module, which, when run, can perform corresponding functions.

[0075] Based on the same inventive concept, in one embodiment, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the drill pipe identification method as described above.

[0076] More details and beneficial effects of this embodiment can be found in the descriptions of the foregoing embodiments, and will not be repeated here.

[0077] Based on the same inventive concept, combined with Figure 11 , Figure 11This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. In one embodiment, this application provides an electronic device 400, including a processor 401, a communication interface 402, a memory 403, and a communication bus 404, wherein the processor, communication interface, and memory communicate with each other through the communication bus; the memory is used to store computer programs; the processor is used to execute the program stored in the memory to implement the method described above. The steps of implementing the drill pipe identification method in any of the above possible implementations can be equivalent to the drill pipe identification system described above. Of course, the processor can also be used to process other data or perform calculations. The electronic device can be a PC, server, terminal, or other devices. More details and beneficial effects of this embodiment can be referred to the relevant descriptions of the foregoing embodiments, which will not be repeated here. It should be understood that these embodiments are only used to illustrate this application and are not intended to limit the scope of this application. Experimental methods in the following embodiments that do not specify specific conditions are generally determined according to national standards. If there is no corresponding national standard, then general international standards, conventional conditions, or conditions recommended by the manufacturer are followed. It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The above descriptions are merely specific embodiments of this disclosure, enabling those skilled in the art to understand or implement this disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A drill pipe identification method, characterized in that, include: Acquire raw video data collected in the earthquake field acquisition project; wherein, the raw video records the dynamic operation process of the drill rod in the field operation scene, including the real-time movement of the drill rod and the interaction information between the operator and the drill rod; The original video is preprocessed; wherein, the preprocessing includes image stabilization. Based on the preprocessed video data, the image change rate of the video data is analyzed, and a stable interval is determined according to the image change rate. Retain the video data corresponding to the time segment to which the stable interval belongs, and remove the video data corresponding to the time segment to which the stable interval belongs; The drill rod is identified and its position is tracked in video data within the stable range using a pre-trained deep learning model.

2. The drill pipe identification method according to claim 1, characterized in that, The image stabilization process includes using ORB technology to stabilize the original video.

3. The drill pipe identification method according to claim 1, characterized in that, The step of analyzing the image change rate of the preprocessed video data and determining the stable interval based on the image change rate includes: Based on the preprocessed video data, the video is divided into multiple regions, and the image change rate of the multiple regions is statistically analyzed in chronological order to generate change rate curves for the multiple regions. The rate of change curve was analyzed using an unsupervised machine learning algorithm; Based on the volatility classification of the curves, groups with low volatility are designated as stable intervals.

4. The drill pipe identification method according to claim 1, characterized in that, The process of identifying and tracking the drill rod's position in video data within the stable interval using a pre-trained deep learning model includes: Using a pre-trained deep learning model, the boundary of the drill rod is outlined in the video data within the stable interval, and the drill rod is identified and its position tracked.

5. The drill pipe identification method according to claim 4, characterized in that, The method further includes: identifying the posture of the operator in the video data.

6. The drill pipe identification method according to claim 5, characterized in that, The method further includes: Extract the time-series data of the drill rod's length and angle changes, and the time-series data of the operator's posture and movements; The time-domain signal corresponding to the time-series data is converted into a frequency-domain signal by Fourier transform, and the periodic characteristics of the drill rod length, angle and the operator's posture are analyzed. Based on the periodicity, the number of drill rods is determined by associating the changes in the length and angle of the drill rod with the synchronicity of the operator's posture and movements.

7. The drill pipe identification method according to any one of claims 1-6, characterized in that, The deep learning model employs data augmentation techniques during the training phase, including random cropping, rotation, and color transformation.

8. The drill pipe identification method according to any one of claims 1-6, characterized in that, The method further includes optimizing the performance of the deep learning model by adjusting its structure or parameters.

9. A drill pipe identification system, characterized in that, include: The first module is used to acquire raw video data collected in the earthquake field acquisition project; wherein, the raw video records the dynamic operation process of the drill rod in the field operation scene, including the real-time movement of the drill rod and the interaction information between the operator and the drill rod; The second module is used to preprocess the original video; wherein, the preprocessing includes image stabilization. The third module is used to analyze the image change rate of the preprocessed video data and determine the stable range based on the image change rate. The fourth module is used to retain the video data corresponding to the time segment to which the stable interval belongs, and to remove the video data corresponding to the time segment outside the stable interval. The fifth module is used to identify and track the position of the drill rod in video data within the stable interval using a pre-trained deep learning model.

10. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor, when executing a program stored in memory, implements the drill pipe identification method according to any one of claims 1 to 8.

Images