Adas data labeling and management method and system
By optimizing the ADAS data annotation tool, efficient frame localization and batch operations are achieved, solving the problems of low efficiency, inaccurate localization, and poor user experience of existing tools, and improving data annotation efficiency and storage management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI GEOMETRICAL PERCEPTION & LEARNING CO LTD
- Filing Date
- 2026-05-25
- Publication Date
- 2026-06-30
AI Technical Summary
Existing ADAS data annotation tools are inefficient, have inaccurate frame localization, poor user experience, and data storage is redundant and has weak correlation.
This paper provides a method for ADAS data annotation and management. By listening to in-vehicle system events and keyboard functions, it realizes standardized processing, batch operation and highlighting of frame data, supports frame number jump and automatic scroll positioning, and optimizes data storage and management.
It improves annotation efficiency by more than 80%, achieves 100% frame positioning accuracy, significantly reduces labor costs, optimizes the interactive experience, and ensures the relevance of data storage.
Smart Images
Figure CN122309733A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automotive electronics and data processing technology, specifically to an ADAS data annotation and management method and system. Background Technology
[0002] Advanced Driver Assistance Systems (ADAS) play a central role in enhancing vehicle driving safety. During the development, testing, and optimization of ADAS, vehicles generate massive amounts of chassis information frames and warning event data. Accurate scene labeling of this data is a crucial prerequisite for analyzing ADAS system performance and training and optimizing perception and decision-making algorithms.
[0003] Currently, most annotation tools for ADAS chassis time-series data borrow from general video annotation software, which has significant technical shortcomings: 1. Low annotation efficiency: Existing tools mainly rely on operators to view and label data frames one by one. Due to the high frequency of ADAS data acquisition (usually at the millisecond level), when faced with driving data that can last for several hours, the single-frame operation method results in a huge workload and extremely long time consumption for annotation.
[0004] 2. Poor frame location capability: On the one hand, existing tools lack a precise frame number input jump mechanism, making it difficult to quickly locate the moment when a specific event occurs; on the other hand, when the system automatically replays or jumps, the visualization list cannot automatically scroll to the current frame position, causing the operator to lose sight of the scene and reducing the continuity of analysis.
[0005] 3. Lack of batch interaction logic: Existing tools usually do not support batch selection of multiple frames triggered by keyboard shortcuts (such as Shift), which makes it impossible to assign labels in batches when processing a series of alarm events. The interactive experience and operation logic fail to match the characteristics of ADAS data continuity.
[0006] 4. Redundant and weak data storage: Existing annotation tools often store all data (including unannotated empty data) indiscriminately, which not only wastes storage resources, but also makes it difficult to establish a solid traceability link with the original vehicle data source (URL) when persisting data.
[0007] Therefore, there is an urgent need for an annotation method and system that can improve the efficiency of ADAS data annotation, optimize the interactive experience, and achieve accurate frame positioning and batch operations. Summary of the Invention
[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide an ADAS data annotation and management method and system, which aims to solve the problems of low efficiency, inaccurate frame positioning, and poor interactive experience in existing ADAS data annotation.
[0009] To achieve the above objectives, the ADAS data annotation and management method and system of the present invention are as follows: The main feature of this ADAS data annotation and management method is that the method includes the following steps: (1) Initialize the system, listen to the frame data change events of the vehicle system and the status of specific function keys on the keyboard, and obtain the preset ADAS alarm tag mapping table and historical tag data from the server; (2) Parse the obtained vehicle chassis information frame data, extract the sequence number and original timestamp of each frame, and convert the original timestamp to a standardized timestamp; traverse each frame data, associate and match it with the historical labeled data based on the frame sequence number, and generate a standardized frame data list containing the frame sequence number, label list and standardized timestamp. (3) Visualize the standardized frame data list and respond to frame click operations; wherein, when no specific function key is detected to be triggered, the clicked frame is recorded as the current frame, the vehicle system is triggered to jump to the corresponding timestamp and highlight the frame; when a specific function key is detected to be continuously triggered and there is a start frame record, the clicked frame is recorded as the end frame, the range of continuous frames defined by the start frame and the end frame is determined, and all frames in the range are highlighted in batches; (4) In response to the tag selection and modification command, the selected target tag is assigned to the currently selected frame object; wherein, if there is a continuous frame range, the target tag is assigned in batch to the tag list of all frames in the range; if there is only the current frame, the target tag is assigned only to the tag list of the current frame. (5) Filter the standardized frame data list, retain the frame data whose tag list is not empty, serialize it into a data string of a preset format, and submit it to the server for persistent storage with the vehicle data source identification information; (6) Receive the save result status returned by the server and generate corresponding operation feedback prompts.
[0010] Preferably, the format conversion of the original timestamp in step (2) specifically involves: Convert the raw large integer timestamp to a string; Determine the string length: If the string length is greater than the preset number of digits threshold, the first part of the string is taken as the integer part and the last part as the decimal part; if the string length is less than or equal to the preset number of digits threshold, the integer part is set to zero and zeros are padded at the beginning of the string to form the decimal part. The integer and fractional parts are combined to form a standardized timestamp in floating-point format.
[0011] Preferably, step (3) determines the range of consecutive frames in the following manner: Record the frame number of the first click as the starting frame number; When the specific function key is detected to be in a continuously triggered state, the frame number of the next click is recorded as the end frame number; The starting frame number and the ending frame number are compared. If the starting frame number is greater than the ending frame number, their values are swapped to form a positive continuous frame range.
[0012] Preferably, step (3) further includes performing automatic scrolling positioning processing: Listen for changes in the current frame; When it is determined that the change in the current frame was not triggered by a manual scrolling operation, retrieve the element corresponding to the current frame in the visualization list; Execute the automatic scrolling command to move the element to a preset position within the visible area.
[0013] Preferably, the method further includes performing frame number jump processing: Obtain the target frame number from the input; The target frame number is compared and verified with the system's preset effective frame range, which is defined by the minimum frame number and the maximum frame number. If the target frame number is less than the minimum frame number, it is automatically corrected to the minimum frame number; if the target frame number is greater than the maximum frame number, it is automatically corrected to the maximum frame number. The system locates the standardized timestamp corresponding to the corrected frame number, triggering the vehicle system to perform data location.
[0014] The ADAS data annotation and management system for implementing the above-described method is characterized in that the system comprises: The data acquisition module is used to acquire frame data containing chassis information from the vehicle system in real time, and parse the frame sequence number and original timestamp; at the same time, it acquires the preset ADAS alarm label mapping table and historical label data from the server. The frame data processing module is used to associate the frame data with the historical labeled data based on the frame sequence number to generate a standardized frame data structure; the standardized frame data structure includes the frame sequence number, the label list, and the standardized timestamp after format conversion. The interactive annotation module provides a human-computer interaction interface, responds to single-frame selection operations and multi-frame range selection operations triggered by specific function key combinations, and provides a label selection interface; the interactive annotation module is further configured to batch assign selected labels to all frame objects within a single frame or multi-frame range. The visualization module is used to render and display the processed frame data in a list format, and to highlight the current frame and the selected frame range differently according to the instructions of the interactive annotation module; the visualization module is also configured to automatically scroll the corresponding list items to a preset position within the visible area when the current frame changes. The data storage module is used to filter the labeled data, extracting only frame data whose label list is not empty, and serializing the filtered data into a preset format for submission to the server for storage; and The monitoring module is used to monitor real-time frame events sent by the vehicle system and the state of specific function keys of the input device to drive the multi-frame selection logic of the interactive annotation module.
[0015] Preferably, the frame data processing module further includes a timestamp conversion unit, used to convert the large integer raw timestamp output by the vehicle system into a standardized timestamp in floating-point format, specifically: Convert the large integer original timestamp into a string; When the string length is greater than 9 characters, the first part is taken as the integer part and the last 9 characters are taken as the decimal part. When the string length is less than or equal to 9 characters, the integer part is set to zero, and the string is padded with zeros at the beginning to 9 characters to form the decimal part; Combine the integer part and the fractional part to form a floating-point number.
[0016] Preferably, the interactive annotation module further includes: The frame number jump unit is used to receive the input target frame number, compare and verify the target frame number with the preset valid frame range; when the target frame number exceeds the valid frame range, it automatically corrects it to the closest boundary frame number; and triggers the vehicle system to locate the standardized timestamp position corresponding to the corrected frame number.
[0017] Preferably, the multi-frame range selection logic of the interactive annotation module is configured as follows: Record the frame number of the first click as the starting frame number; When the monitoring module detects that a specific function key is in a continuously triggered state, it records the frame number of the next click as the end frame number. If the starting frame number is greater than the ending frame number, then the values of the two are swapped to determine the range of consecutive frames.
[0018] Preferably, the visualization module further includes: The scroll control unit listens for changes in the current frame and, when a non-manual operation is detected, calls the browser's native scrolling interface to move the list item element corresponding to the current frame to the top of the visible area.
[0019] The ADAS data annotation and management method and system of this invention have been applied to an in-vehicle ADAS testing platform. Actual testing has shown that annotation efficiency has been improved by more than 80%, frame positioning accuracy has reached 100%, significantly reducing the labor cost of ADAS data annotation and improving data analysis efficiency. Attached Figure Description
[0020] Figure 1 This is a flowchart of the ADAS data annotation and management method of the present invention.
[0021] Figure 2 This is a module architecture diagram of the ADAS data annotation and management system of the present invention.
[0022] Figure 3 This is a flow diagram of the mcap file and labeled JSON data in a specific embodiment of the present invention.
[0023] Figure 4 This is a schematic diagram visually demonstrating the operation interface in a specific embodiment of the present invention. Detailed Implementation
[0024] To more clearly describe the technical content of the present invention, the following description is provided in conjunction with specific embodiments.
[0025] Before describing the embodiments of the present invention in detail, it should be noted that, in the following, the terms “comprising,” “including,” or any other variations 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 inherent to such process, method, article, or apparatus.
[0026] Please see Figure 1 As shown, this ADAS data annotation and management method includes the following steps: (1) Initialize the system, listen to the frame data change events of the vehicle system and the status of specific function keys on the keyboard, and obtain the preset ADAS alarm tag mapping table and historical tag data from the server; (2) Parse the obtained vehicle chassis information frame data, extract the sequence number and original timestamp of each frame, and convert the original timestamp to a standardized timestamp; traverse each frame data, associate and match it with the historical labeled data based on the frame sequence number, and generate a standardized frame data list containing the frame sequence number, label list and standardized timestamp. (3) Visualize the standardized frame data list and respond to frame click operations; wherein, when no specific function key is detected to be triggered, the clicked frame is recorded as the current frame, the vehicle system is triggered to jump to the corresponding timestamp and highlight the frame; when a specific function key is detected to be continuously triggered and there is a start frame record, the clicked frame is recorded as the end frame, the range of continuous frames defined by the start frame and the end frame is determined, and all frames in the range are highlighted in batches; (4) In response to the tag selection and modification command, the selected target tag is assigned to the currently selected frame object; wherein, if there is a continuous frame range, the target tag is assigned in batch to the tag list of all frames in the range; if there is only the current frame, the target tag is assigned only to the tag list of the current frame. (5) Filter the standardized frame data list, retain the frame data whose tag list is not empty, serialize it into a data string of a preset format, and submit it to the server for persistent storage with the vehicle data source identification information; (6) Receive the save result status returned by the server and generate corresponding operation feedback prompts.
[0027] In a preferred embodiment of the present invention, the format conversion of the original timestamp in step (2) specifically involves: Convert the raw large integer timestamp to a string; Determine the string length: If the string length is greater than the preset number of digits threshold, the first part of the string is taken as the integer part and the last part as the decimal part; if the string length is less than or equal to the preset number of digits threshold, the integer part is set to zero and zeros are padded at the beginning of the string to form the decimal part. The integer and fractional parts are combined to form a standardized timestamp in floating-point format.
[0028] In a preferred embodiment of the present invention, step (3) determines the range of consecutive frames in the following manner: Record the frame number of the first click as the starting frame number; When the specific function key is detected to be in a continuously triggered state, the frame number of the next click is recorded as the end frame number; The starting frame number and the ending frame number are compared. If the starting frame number is greater than the ending frame number, their values are swapped to form a positive continuous frame range.
[0029] In a preferred embodiment of the present invention, step (3) further includes performing automatic rolling positioning processing: Listen for changes in the current frame; When it is determined that the change in the current frame was not triggered by a manual scrolling operation, retrieve the element corresponding to the current frame in the visualization list; Execute the automatic scrolling command to move the element to a preset position within the visible area.
[0030] In a preferred embodiment of the present invention, the method further includes performing frame number jump processing: Obtain the target frame number from the input; The target frame number is compared and verified with the system's preset effective frame range, which is defined by the minimum frame number and the maximum frame number. If the target frame number is less than the minimum frame number, it is automatically corrected to the minimum frame number; if the target frame number is greater than the maximum frame number, it is automatically corrected to the maximum frame number. The system locates the standardized timestamp corresponding to the corrected frame number, triggering the vehicle system to perform data location.
[0031] In a preferred embodiment of the present invention, the keyboard-specific function key mentioned above is the Shift key in practical applications.
[0032] The implementation of this ADAS data annotation and management method will be described in detail below with reference to specific embodiments: System initialization: Listen for events such as currentFrame and allFrames in the vehicle system and subscribe to the chassis_info topic; Retrieve ADAS alarm tag mapping table (e.g., 8007:FCTA Left, 8107:FCTA Right, etc.) and historical tag JSON data from the server; Listen for the pressed / released state of the Shift key on the keyboard and maintain the isShiftPressed state.
[0033] 2. Frame data processing: Parse the chassis_info frame data and extract the frame (frame number) and time_meas (large integer timestamp). Call the bigIntToNumber function to convert time_meas into a decimal timestamp; Associate with historical labeled data: If records with the same frame exist in the historical data, reuse their labels; otherwise, initialize labels to an empty array and generate a handleData list.
[0034] 3. Visualize the list of frame data and listen for frame click events: When the Shift key is not held down, the clicked frame number is recorded as the current frame (currentFrame), and the system jumps to the vehicle data timestamp corresponding to that frame, highlights the frame, and clears the frame range. If you hold down the Shift key and the start frame number has been recorded, record the end frame number. If the start frame number is greater than the end frame number, swap the order to form a valid frame range and highlight all frames within that range.
[0035] 4. Frame selection and assignment: Select the target label (such as 8007, 8107, etc.) from the label dropdown menu, and click the "Modify" button to trigger the label assignment: When a valid frame range exists, update the labels of all frames within the range to the selected label; If only the current frame exists, update the label of the current frame to the selected label.
[0036] 5. Data storage: Filter frames in the handleData list whose labels are not empty and serialize them into JSON strings; Submit the vehicle data URL parameter to the server's / saveJson interface; Receive the server response and display "Modification successful" or "Modification failed" in the prompt box.
[0037] 6. Frame number jump: Enter the target frame number in the input box and check if the frame number is within the valid range (minimum frame to maximum frame). If the data exceeds the specified range, it will automatically correct to the boundary frame, locate the chassis data timestamp of the corresponding frame, and trigger the vehicle system to play the positioning.
[0038] Please see Figure 2 As shown, the ADAS data annotation and management system includes: The data acquisition module is used to acquire frame data containing chassis information from the vehicle system in real time, and parse the frame sequence number and original timestamp; at the same time, it acquires the preset ADAS alarm label mapping table and historical label data from the server. The frame data processing module is used to associate the frame data with the historical labeled data based on the frame sequence number to generate a standardized frame data structure; the standardized frame data structure includes the frame sequence number, the label list, and the standardized timestamp after format conversion. The interactive annotation module provides a human-computer interaction interface, responds to single-frame selection operations and multi-frame range selection operations triggered by specific function key combinations, and provides a label selection interface; the interactive annotation module is further configured to batch assign selected labels to all frame objects within a single frame or multi-frame range. The visualization module is used to render and display the processed frame data in a list format, and to highlight the current frame and the selected frame range differently according to the instructions of the interactive annotation module; the visualization module is also configured to automatically scroll the corresponding list items to a preset position within the visible area when the current frame changes. The data storage module is used to filter the labeled data, extracting only frame data whose label list is not empty, and serializing the filtered data into a preset format for submission to the server for storage; and The monitoring module is used to monitor real-time frame events sent by the vehicle system and the state of specific function keys of the input device to drive the multi-frame selection logic of the interactive annotation module.
[0039] In a preferred embodiment of the present invention, the large integer timestamp conversion method in the frame data processing module is as follows: the large integer is converted into a string; if the string length is greater than 9 characters, the first N-9 characters are truncated to be the integer part and the last 9 characters are the decimal part; if the string length is ≤ 9 characters, zeros are padded to be the decimal part, and the integer part is 0, ultimately converting it into a floating-point number. The specific processing procedure is as follows: Application scenario: Large integer timestamps output by in-vehicle systems cannot be directly read and calculated, and need to be converted to standard decimal format.
[0040] Input: Vehicle-mounted raw large integer timestamp.
[0041] Processing steps: 1. Convert the large integer timestamp to the string str_time; 2. Get the string length len; 3. Length determination and segmentation: If len > 9: the integer part = the first len - 9 digits, and the fractional part = the last 9 digits; If len ≤ 9: the integer part = 0, the fractional part = padded with zeros on the left to 9 digits; 4. Combine into a floating-point format timestamp.
[0042] Conversion formula: timestamp_float = int (str_time [0:-9]) + int (str_time [-9:]) / 1000000000 Example: 1720001234567890000 → 1720001234.567890000.
[0043] As a preferred embodiment of the present invention, in the interactive annotation module, the multi-frame range selection logic is as follows: When clicking on a frame for the first time, record the starting frame number. While holding down the Shift key and clicking on a frame again, record the ending frame number to form a continuous frame range. The continuous multi-frame quick selection is achieved through the Shift key, replacing the traditional single-frame click. The specific processing process is as follows: Status variables: start_frame (starting frame), end_frame (ending frame), is_shift_pressed (Shift key pressed status) Execution logic: 1. Initialization: start_frame = -1, end_frame = -1, is_shift_pressed = false; 2. Button monitoring: Press Shift → is_shift_pressed = true; Release Shift → is_shift_pressed = false; 3. Frame click operation: Without pressing Shift: Set start_frame = the current frame number, clear the frame range, and only select a single frame; While holding down Shift and start_frame ≠ -1: Set end_frame = the current frame number; If start_frame>end_frame: Swap their values to form a legal continuous frame interval; 4. Batch assignment: Uniformly assign the selected label to all frames within [start_frame, end_frame].
[0044] As a preferred embodiment of the present invention, in the interactive annotation module, the frame number jump verification is mainly used to prevent positioning failure caused by entering an invalid frame number and achieve accurate frame jumping. The logical processing steps are as follows: 1. Obtain the valid frame range of the system: min_frame (minimum frame) ~ max_frame (maximum frame); 2. Obtain the target frame target_frame input by the user; 3. Legality verification and correction: target_frame < min_frame → Automatically correct to min_frame; target_frame > max_frame → Automatically correct to max_frame; Legal frame number → Use directly; 4. Jump to execution: Locate the corresponding frame timestamp and trigger the vehicle system to play the location.
[0045] In a preferred embodiment of the present invention, the frame highlighting rule in the visualization module is as follows: the current frame and frames within the frame range adopt differentiated styles (such as background color highlighting), while unselected frames adopt the default style; the automatic scrolling positioning logic is as follows: when the current frame changes and there is no manual click operation, the DOM element of the current frame is automatically scrolled to the top of the view. The specific execution logic is as follows: 1. Monitor system frame change events; 2. Determine if there is no manual mouse / keyboard operation; 3. Get the DOM element corresponding to the current frame; 4. Perform scrolling: scrollIntoView({top: 0, behavior: 'smooth'}); 5. Style differentiation: Different highlight styles are used for the current frame, the selected frame range, and ordinary frames.
[0046] In a preferred embodiment of the present invention, the filtering logic in the data storage module is as follows: only frame data with a label list length greater than 0 is retained, and data with empty labels is directly removed to avoid unlabeled data occupying storage resources.
[0047] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.
[0048] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution device.
[0049] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
[0050] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.
[0051] In the description of this specification, references to terms such as "an embodiment," "some embodiments," "example," "specific example," or "embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0052] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.
[0053] The ADAS data annotation and management method and system of this invention have the following significant advantages compared with the prior art: 1. Improve annotation efficiency: Supports multi-frame range selection and batch label assignment triggered by the Shift key, replacing the traditional single-frame annotation method and significantly reducing the number of operations; 2. Precise frame positioning: Supports frame number input for jumping, and automatically scrolls to the visual position when the current frame changes, solving the problem of inaccurate frame positioning; 3. Optimized user experience: Frame selection and tag selection are linked, the selected frame and the current frame are highlighted, and the operation feedback is intuitive; 4. Convenient data storage: After labeling, unlabeled data is automatically filtered and submitted to the server for storage, and the association with the vehicle data URL ensures data traceability; 5. Strong compatibility: Adapts to the large integer timestamp format of in-vehicle systems, converts it to an easy-to-read decimal format, and improves data readability.
[0054] In this specification, the invention has been described with reference to specific embodiments thereof. However, it will be apparent that various modifications and variations can be made without departing from the spirit and scope of the invention. Therefore, the specification and drawings should be considered illustrative rather than restrictive.
Claims
1. A method for ADAS data annotation and management, characterized in that, The method includes the following steps: (1) Initialize the system, listen to the frame data change events of the vehicle system and the status of specific function keys on the keyboard, and obtain the preset ADAS alarm tag mapping table and historical tag data from the server; (2) Parse the obtained vehicle chassis information frame data, extract the sequence number and original timestamp of each frame, and convert the original timestamp to a standardized timestamp; traverse each frame data, associate and match it with the historical labeled data based on the frame sequence number, and generate a standardized frame data list containing the frame sequence number, label list and standardized timestamp. (3) Visualize the standardized frame data list and respond to frame click operations; wherein, when no specific function key is detected to be triggered, the clicked frame is recorded as the current frame, the vehicle system is triggered to jump to the corresponding timestamp and highlight the frame; when a specific function key is detected to be continuously triggered and there is a start frame record, the clicked frame is recorded as the end frame, the range of continuous frames defined by the start frame and the end frame is determined, and all frames in the range are highlighted in batches; (4) In response to the tag selection and modification command, the selected target tag is assigned to the currently selected frame object; wherein, if there is a continuous frame range, the target tag is assigned in batch to the tag list of all frames in the range; if there is only the current frame, the target tag is assigned only to the tag list of the current frame. (5) Filter the standardized frame data list, retain the frame data whose tag list is not empty, serialize it into a data string of a preset format, and submit it to the server for persistent storage with the vehicle data source identification information; (6) Receive the save result status returned by the server and generate corresponding operation feedback prompts.
2. The ADAS data annotation and management method according to claim 1, characterized in that, The specific steps for format conversion of the original timestamp in step (2) are as follows: Convert the raw large integer timestamp to a string; Determine the string length: If the string length is greater than the preset number of digits threshold, then extract the first part of the string as the integer part and the second part as the decimal part; If the string length is less than or equal to the preset number of digits threshold, the integer part is set to zero, and zeros are padded at the beginning of the string to form the decimal part; The integer and fractional parts are combined to form a standardized timestamp in floating-point format.
3. The ADAS data annotation and management method according to claim 1, characterized in that, Step (3) determines the range of consecutive frames in the following manner: Record the frame number of the first click as the starting frame number; When the specific function key is detected to be in a continuously triggered state, the frame number of the next click is recorded as the end frame number; The starting frame number and the ending frame number are compared. If the starting frame number is greater than the ending frame number, their values are swapped to form a positive continuous frame range.
4. The ADAS data annotation and management method according to claim 1, characterized in that, Step (3) further includes performing automatic scrolling positioning processing: Listen for changes in the current frame; When it is determined that the change in the current frame was not triggered by a manual scrolling operation, retrieve the element corresponding to the current frame in the visualization list; Execute the automatic scrolling command to move the element to a preset position within the visible area.
5. The ADAS data annotation and management method according to claim 1, characterized in that, The method also includes frame number jump processing: Obtain the target frame number from the input; The target frame number is compared and verified with the system's preset effective frame range, which is defined by the minimum frame number and the maximum frame number. If the target frame number is less than the minimum frame number, it is automatically corrected to the minimum frame number; if the target frame number is greater than the maximum frame number, it is automatically corrected to the maximum frame number. The system locates the standardized timestamp corresponding to the corrected frame number, triggering the vehicle system to perform data location.
6. An ADAS data annotation and management system for implementing the method of any one of claims 1 to 5, characterized in that, The system includes: The data acquisition module is used to acquire frame data containing chassis information from the vehicle system in real time, and parse the frame sequence number and original timestamp; at the same time, it acquires the preset ADAS alarm label mapping table and historical label data from the server. The frame data processing module is used to associate the frame data with the historical labeled data based on the frame sequence number to generate a standardized frame data structure; the standardized frame data structure includes the frame sequence number, the label list, and the standardized timestamp after format conversion. The interactive annotation module provides a human-computer interaction interface, responds to single-frame selection operations and multi-frame range selection operations triggered by specific function key combinations, and provides a label selection interface; the interactive annotation module is further configured to batch assign selected labels to all frame objects within a single frame or multi-frame range. The visualization module is used to render and display the processed frame data in a list format, and to highlight the current frame and the selected frame range differently according to the instructions of the interactive annotation module; the visualization module is also configured to automatically scroll the corresponding list items to a preset position within the visible area when the current frame changes. The data storage module is used to filter the labeled data, extracting only frame data whose label list is not empty, and serializing the filtered data into a preset format for submission to the server for storage; and The monitoring module is used to monitor real-time frame events sent by the vehicle system and the state of specific function keys of the input device to drive the multi-frame selection logic of the interactive annotation module.
7. The ADAS data annotation and management system according to claim 6, characterized in that, The frame data processing module further includes a timestamp conversion unit, used to convert the large integer raw timestamps output by the vehicle system into standardized timestamps in floating-point format, specifically: Convert the large integer original timestamp into a string; When the string length is greater than 9 characters, the first part is taken as the integer part and the last 9 characters are taken as the decimal part. When the string length is less than or equal to 9 characters, the integer part is set to zero, and the string is padded with zeros at the beginning to 9 characters to form the decimal part; Combine the integer part and the fractional part to form a floating-point number.
8. The ADAS data annotation and management system according to claim 6, characterized in that, The interactive annotation module also includes: The frame number jump unit is used to receive the input target frame number, compare and verify the target frame number with the preset valid frame range; when the target frame number exceeds the valid frame range, it automatically corrects it to the closest boundary frame number; and triggers the vehicle system to locate the standardized timestamp position corresponding to the corrected frame number.
9. The ADAS data annotation and management system according to claim 6, characterized in that, The multi-frame range selection logic of the interactive annotation module is configured as follows: Record the frame number of the first click as the starting frame number; When the monitoring module detects that a specific function key is in a continuously triggered state, it records the frame number of the next click as the end frame number. If the starting frame number is greater than the ending frame number, then the values of the two are swapped to determine the range of consecutive frames.
10. The ADAS data annotation and management system according to claim 6, characterized in that, The visualization module also includes: The scroll control unit listens for changes in the current frame and, when a non-manual operation is detected, calls the browser's native scrolling interface to move the list item element corresponding to the current frame to the top of the visible area.