A method and system for annotating signal symbols

Abstract

This invention belongs to the field of data annotation technology and discloses a signal graph annotation method and system. The method includes: Step 1: Reading signal data and generating a high-resolution graph and a synthetic graph; Step 2: Inputting the generated high-resolution graph into a signal detection model to generate graph pre-annotation results; Step 3: Judging the correctness of the graph pre-annotation results by referring to the synthetic graph, correcting erroneous graph pre-annotation results, and obtaining the final graph annotation results; Step 4: Saving the graph data annotation file based on the final graph annotation results. This invention automatically generates graph pre-annotation results by inputting the generated high-resolution graph into a signal detection model, which can significantly reduce the workload of manual annotation. Users can judge the correctness of the graph pre-annotation results by referring to the synthetic graph and correct erroneous graph pre-annotation results, thereby improving the accuracy of annotation.

Description

Technical Field

[0001] This invention relates to the field of data annotation technology, and in particular to a method and system for annotating signal symbols. Background Technology

[0002] In the field of radio signal reception and detection, speech maps are used to visually display the characteristics of signals. Through speech maps, it is possible to determine whether there is a signal in the image, what type of signal it is, and the location of the signal. They are widely used in practical signal target detection engineering.

[0003] In recent years, with the rapid development of artificial intelligence (AI) technology, AI has been widely applied in various fields, including object detection, image recognition, and image classification. Annotated image data provides training references for machine learning and is crucial in training neural network models. Existing image object annotation systems typically annotate single images, resulting in low annotation efficiency and time-consuming, labor-intensive work. Summary of the Invention

[0004] This invention addresses the problems of large training data volumes and time-consuming annotation of signal graphs by providing a signal graph annotation method and system. The system can call a signal detection model to generate pre-annotated graph results, then compare these results with a synthesized graph to determine their correctness, correcting any erroneous annotations. Furthermore, it can annotate multiple graphs simultaneously, significantly improving the speed and efficiency of data annotation. The final annotated data can be used for signal target detection and the training and optimization of the signal detection model (a neural network model).

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] This invention proposes a method for annotating signal symbols, comprising:

[0007] Step 1: Read signal data and generate high-resolution and synthetic graphs;

[0008] Step 2: Input the generated high-resolution graph into the signal detection model to generate graph pre-annotation results;

[0009] Step 3: Refer to the synthesized speech graph to determine the correctness of the speech graph pre-annotation results, correct any incorrect speech graph pre-annotation results, and obtain the final speech graph annotation results;

[0010] Step 4: Save the graph data annotation file based on the final graph annotation results.

[0011] Further, step 1 includes:

[0012] Step 1.1: Read the speech graph data files to be labeled. Read multiple high-resolution speech graph data files and multiple synthetic speech graph data files each time, and the number of high-resolution speech graph data files and synthetic speech graph data files read each time is the same.

[0013] Step 1.2: Obtain multiple floating-point data from each high-resolution image data file, form a one-dimensional list, and transform it into a two-dimensional matrix through matrix operations. Then, display the two-dimensional matrix data as an image to obtain the high-resolution image.

[0014] Step 1.3: Construct complex data based on the composition of the synthesized graph data file, perform a fast Fourier transform on the constructed complex data to obtain the spectral energy distribution of the signal, derive the synthesized graph based on the spectral energy distribution, and display it.

[0015] Furthermore, in step 1.3, complex data is constructed in the following manner:

[0016] X = I + jQ

[0017] Where X is complex data, I and Q are two streams of data from the composite graph data file, and j is the complex unit.

[0018] Furthermore, the high-resolution speech graph and synthesized speech graph... Figure One One-to-one correspondence.

[0019] Furthermore, the pre-annotation results of the speech graph include the signal region, the signal type within the region, and the probability of belonging to that signal type.

[0020] Further, step 4 includes:

[0021] Based on the final image annotation results, the labels and image data are stored as .dat files, with multiple sets of labels and image data saved each time; each set contains one set of label data and one image data; all data are arranged in a one-dimensional manner, and each save uses an append storage mode.

[0022] Another aspect of the present invention proposes a signal symbol annotation system, comprising:

[0023] The data reading module is used to read signal data and generate high-resolution and synthetic graphs.

[0024] The graph pre-annotation module is used to generate graph pre-annotation results from the high-resolution graph input signal detection model;

[0025] The pre-annotation result correction module is used to judge the correctness of the pre-annotation results of the synthesized graph by referring to the synthesized graph, correct the incorrect pre-annotation results of the graph, and obtain the final graph annotation results;

[0026] The label data storage module is used to save the graph data annotation file based on the final graph annotation results.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] This invention can simultaneously display high-resolution graphs and synthetic graphs with a one-to-one correspondence. Users can annotate the high-resolution graph by referring to the signal results of the synthetic graph. Specifically, this invention automatically generates graph pre-annotation results from the input signal detection model of the generated high-resolution graph. Users can judge the correctness of the graph pre-annotation results by referring to the synthetic graph and correct erroneous graph pre-annotation results, thereby improving the accuracy of annotation and significantly reducing the workload of manual annotation. Attached Figure Description

[0029] Figure 1 This is a basic flowchart of a signal symbol annotation method according to an embodiment of the present invention;

[0030] Figure 2 This is a schematic diagram of the data storage structure provided in an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram of a tag data structure provided in an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of a signal and icon annotation system architecture according to an embodiment of the present invention;

[0033] Figure 5 The signal symbol annotation system interface provided in this embodiment of the invention. Detailed Implementation

[0034] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments:

[0035] like Figure 1 As shown, a signal icon annotation method includes:

[0036] S101: Read signal data and generate high-resolution and synthetic graphs;

[0037] S102: Input the generated high-resolution graph into the signal detection model (a neural network model can be used, specifically the YOLOv5s neural network model) to generate graph pre-annotation results;

[0038] S103: Refer to the synthesized speech graph to determine the correctness of the speech graph pre-annotation results, correct the incorrect speech graph pre-annotation results, and obtain the final speech graph annotation results;

[0039] S104: Save the graph data annotation file based on the final graph annotation results.

[0040] Further, S101 includes:

[0041] Step S101.1: Read the speech graph data files to be labeled. Each time, read multiple high-resolution speech graph data files and multiple synthetic speech graph data files, and the number of high-resolution speech graph data files and synthetic speech graph data files read each time is the same.

[0042] Step S101.2: Obtain multiple floating-point data from each high-resolution image data file to form a one-dimensional list, and transform it into a two-dimensional matrix through matrix operations. Then, display the two-dimensional matrix data as an image to obtain the high-resolution image.

[0043] Step S101.3: Construct complex data based on the composition of the synthesized graph data file, perform a fast Fourier transform on the constructed complex data to obtain the spectral energy distribution of the signal, derive the synthesized graph based on the spectral energy distribution, and display it.

[0044] In one specific implementation method, S101 reads 10 high-resolution graph data files and 10 synthesized graph data files each time. Each high-resolution graph data file can acquire 10,000 floating-point data points, forming a one-dimensional list. This list is then transformed into an 80*125 two-dimensional matrix through matrix operations. A drawing command is then used to display the two-dimensional matrix data as an image on the system interface. Each synthesized graph data file contains I and Q channels. Complex data X is constructed using the I and Q channels according to Formula 1. A Fast Fourier Transform is performed on the constructed complex data X to obtain the signal's spectral energy distribution. A synthesized graph is then drawn based on the spectral energy distribution and displayed on the system interface. The interface simultaneously displays 10 high-resolution graphs and 10 synthesized graphs (specifically, direction-finding graphs). Figure One One-to-one correspondence.

[0045] Formula 1: X=I+jQ

[0046] In the formula: j is the complex unit.

[0047] Further, S102 includes: according to the signal graph generated in S101, each high-resolution graph data is sequentially sent to the signal detection model to obtain pre-labeling results and displayed on the interface. The pre-labeling result information includes the signal region, the signal type within the region, and the probability of belonging to that signal type.

[0048] Furthermore, S103 includes: the user judging the correctness of the pre-annotation result; if the pre-annotation result is correct, the user can directly adopt it; if the pre-annotation result is incorrect, the user can modify the signal location area and signal type, i.e., the signal area and the signal type within the area, to obtain the final text annotation result.

[0049] Further, step S104 includes: storing the labels and graph data into a .dat file based on the final graph annotation results. The stored .dat file can be further used for signal target detection or training and optimization of the signal detection model. The storage format is as follows: Figure 2 As shown, each time 10 sets of label and graph data are saved, a total of 100,180 numbers are recorded; each set of label and graph data contains 1 set of label data and 1 graph, totaling 10,018 numbers (T1); each set of label data contains 18 numbers, which can represent 3 labeled targets, each target is represented by 6 numbers, such as... Figure 3 As shown, B1 represents the signal type, B1-B5 represent the regions where the signal appears, and B6 is the confidence level of the automatically labeled information; each image contains 10,000 numbers; all data are arranged in a one-dimensional manner, and each save uses an append-only storage mode.

[0050] This invention automatically generates pre-annotation results for the high-resolution graph input signal detection model. Users can refer to the synthesized graph to judge the correctness of the pre-annotation results and correct erroneous pre-annotation results, thereby improving the accuracy of annotation and significantly reducing the workload of manual annotation.

[0051] Based on the above embodiments, such as Figure 4 As shown, the present invention also proposes a signal symbol annotation system, comprising:

[0052] The data reading module is used to read signal data and generate high-resolution and synthetic graphs.

[0053] The graph pre-annotation module is used to generate graph pre-annotation results from the high-resolution graph input signal detection model;

[0054] The pre-annotation result correction module is used to judge the correctness of the pre-annotation results of the synthesized graph by referring to the synthesized graph, correct the incorrect pre-annotation results of the graph, and obtain the final graph annotation results;

[0055] The label data storage module is used to save the graph data annotation file based on the final graph annotation results.

[0056] The specific functions of each module correspond to the functions of each step in a signal symbol annotation method, and will not be elaborated here.

[0057] As one possible implementation method, the interface of the signal symbol annotation system of the present invention is as follows: Figure 5 As shown.

[0058] Based on the original signal data file generated by the business system, the signal graph data annotation system of this invention is used for data annotation. First, the original signal data file is read, then 10 high-resolution graph (signal) data files and 10 composite graph (signal) data files are read each time. The composite graph serves as a reference graph for signal annotation, along with the high-resolution graphs. Figure One One-to-one correspondence; the system includes an initial detection model that can generate pre-annotation results for the speech graph. Users can judge the correctness of the pre-detection results based on the displayed speech graph and modify the incorrect information in the pre-annotation. After completion, the system saves the labels and speech graph data. It can complete the annotation of 10 speech graphs at one time. Combined with the reference of the synthesized speech graph, it ensures both the efficiency of the annotation execution and the accuracy of the annotation information.

[0059] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for annotating signal symbols, characterized in that, include: Step 1: Read signal data and generate high-resolution and synthetic graphs; Step 2: Input the generated high-resolution graph into the signal detection model to generate graph pre-annotation results; Step 3: Refer to the synthesized speech graph to determine the correctness of the speech graph pre-annotation results, correct any incorrect speech graph pre-annotation results, and obtain the final speech graph annotation results; Step 4: Save the graph data annotation file based on the final graph annotation results; Step 1 includes: Step 1.1: Read the speech graph data files to be labeled. Read multiple high-resolution speech graph data files and multiple synthetic speech graph data files each time, and the number of high-resolution speech graph data files and synthetic speech graph data files read each time is the same. Step 1.2: Obtain multiple floating-point data from each high-resolution image data file, form a one-dimensional list, and transform it into a two-dimensional matrix through matrix operations. Then, display the two-dimensional matrix data as an image to obtain the high-resolution image. Step 1.3: Construct complex data based on the composition of the synthesized graph data file, perform a fast Fourier transform on the constructed complex data to obtain the spectral energy distribution of the signal, derive the synthesized graph based on the spectral energy distribution, and display it; In step 1.3, complex data is constructed in the following manner: X=I+jQ Where X is complex data, I and Q are two streams of data from the composite graph data file, and j is the complex unit.

2. The signal and icon annotation method according to claim 1, characterized in that, The high-resolution speech graph and the synthesized speech graph are in one-to-one correspondence.

3. The signal and icon annotation method according to claim 1, characterized in that, The pre-annotation results of the speech graph include the signal region, the signal type within the region, and the probability of belonging to that signal type.

4. The signal symbol annotation method according to claim 3, characterized in that, Step 4 includes: Based on the final image annotation results, the labels and image data are stored as .dat files, with multiple sets of labels and image data saved each time; each set contains one set of label data and one image data; all data are arranged in a one-dimensional manner, and each save uses an append storage mode.

5. A signal symbol annotation system, characterized in that, include: The data reading module is used to read signal data and generate high-resolution and synthetic graphs. The graph pre-annotation module is used to generate graph pre-annotation results from the high-resolution graph input signal detection model; The pre-annotation result correction module is used to judge the correctness of the pre-annotation results of the synthesized graph by referring to the synthesized graph, correct the incorrect pre-annotation results of the graph, and obtain the final graph annotation results; The tag data storage module is used to save the graph data annotation file according to the final graph annotation results; The data reading module is specifically used for: Read the image data file to be labeled. Each time, read multiple high-resolution image data files and multiple synthetic image data files, and the number of high-resolution image data files and synthetic image data files read each time is the same. Multiple floating-point data are obtained from each high-resolution image data file to form a one-dimensional list. Then, matrix operations are used to transform it into a two-dimensional matrix. Finally, the two-dimensional matrix data is displayed as an image to obtain the high-resolution image. Complex data is constructed based on the composition of the synthesized graph data file. A fast Fourier transform is performed on the constructed complex data to obtain the spectral energy distribution of the signal. The synthesized graph is then derived from the spectral energy distribution and displayed. Construct complex numbers as follows: X=I+jQ Where X is complex data, I and Q are two streams of data from the composite graph data file, and j is the complex unit.

Images