CN116069743A - Fluid data compression method based on time sequence characteristics - Google Patents
Fluid data compression method based on time sequence characteristics Download PDFInfo
- Publication number
- CN116069743A CN116069743A CN202310200041.4A CN202310200041A CN116069743A CN 116069743 A CN116069743 A CN 116069743A CN 202310200041 A CN202310200041 A CN 202310200041A CN 116069743 A CN116069743 A CN 116069743A
- Authority
- CN
- China
- Prior art keywords
- data
- point
- weight
- value
- vortex
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 239000012530 fluid Substances 0.000 title claims abstract description 28
- 238000013144 data compression Methods 0.000 title claims abstract description 24
- 238000013138 pruning Methods 0.000 claims abstract description 13
- 238000001514 detection method Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000000605 extraction Methods 0.000 claims description 8
- FGUUSXIOTUKUDN-IBGZPJMESA-N C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 Chemical compound C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 FGUUSXIOTUKUDN-IBGZPJMESA-N 0.000 claims description 3
- 238000003491 array Methods 0.000 claims description 3
- 238000013139 quantization Methods 0.000 abstract description 4
- 238000013500 data storage Methods 0.000 abstract description 3
- 230000005540 biological transmission Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 abstract description 2
- 238000007906 compression Methods 0.000 description 3
- 230000006835 compression Effects 0.000 description 3
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/17—Details of further file system functions
- G06F16/174—Redundancy elimination performed by the file system
- G06F16/1744—Redundancy elimination performed by the file system using compression, e.g. sparse files
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Algebra (AREA)
- Software Systems (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention relates to the technical field of data compression, in particular to a fluid data compression method based on time sequence characteristics, which comprises the following steps: (1) Turbine feature detection, namely processing calculated fluid data through a vortex detection algorithm; (2) pruning a single time step data; (3) time sequential data merging and quantization. The invention focuses on the timeliness of data use, utilizes the timeliness of time sequence data, further solves the redundancy of the data, improves the transmission speed of the data, saves the data storage space, achieves the effect superior to other data compression, and is suitable for data compression with time sequence.
Description
Technical Field
The invention relates to the technical field of data compression, in particular to a fluid data compression method based on time sequence characteristics.
Background
The three-dimensional fluid data has very large data volume, high density and serious data redundancy problem. Based on the above, under the condition that the model accuracy is kept as low as possible in practical application, it is necessary to properly perform compression storage on three-dimensional fluid data. In the existing data compression technology, the main method is as follows:
for the representation method of the time series, main characteristics of the time series are extracted, and the time series is transformed into a low-dimensional space. Common methods for time-series data compression are Singular Value Decomposition (SVD), piecewise Linear Representation (PLR), and the like, which cannot be applied to three-dimensional stream data although some progress has been made in data compression.
For three-dimensional point cloud data, according to the estimated normal vector, the three-dimensional point number is reduced, and the method is applied to uniform compression of the point cloud data. Chen Xijiang et al propose a point cloud compression method based on normal vector included angle information entropy, wherein more points are reserved in a region with rich characteristics, and a few points are reserved in a region with smaller information entropy. Although the three-dimensional point cloud data can compress the three-dimensional fluid data to a certain extent, the three-dimensional point cloud data ignores the time sequence characteristics of the three-dimensional point cloud data and does not utilize the time sequence characteristics of the data, so that the data are compressed by utilizing the time sequence characteristics of the three-dimensional fluid data, and the problems of large data quantity and serious data redundancy are solved.
Disclosure of Invention
In order to solve the problems, the invention provides a fluid data compression method based on time sequence characteristics, which utilizes the attribute of vortex characteristics and the timeliness of data to compress vortex data.
The invention provides the following technical scheme: a method of fluid data compression based on timing characteristics, comprising the steps of: (1) Turbine feature detection, namely processing calculated fluid data through a vortex detection algorithm; (2) pruning a single time step data; (3) time sequential data merging and quantization.
In step (1), use is made ofCriteria, Q criteria, < >>The method combines the methods of vortex quantity and the like to realize vortex characteristic extraction,
criteria: the eigenvalues of the velocity tensor J are used to determine the local streamline characteristics that follow a point in a certain reference frame, at the center of the vortex, J should have a relatively complex eigenvalue, J eigenvalue, X satisfies the following equation:
wherein, the liquid crystal display device comprises a liquid crystal display device,
where tr (J) represents the trace of matrix J and det (J) represents the determinant of matrix J;
j has complex eigenvalues when the following equation holds:
q criterion: the second component Q of the velocity gradient tensor J is positive and the pressure value is here smaller than the pressure value of the surrounding points, the second component Q being defined as,
Is->A centered characteristic value of the three characteristic values of (a) if a coordinate point in the fluid data corresponds to +.>If the condition of (2) indicates that the point is a point in the vortex;
vorticity: the number of rings, which generally describe the speed of rotation and the size of the range of rotation, marks the rotation of the fluid,
will beCriteria, Q criteria, < >>The method and the vortex quantity and other methods give a certain weight proportion to the obtained value, a 32-bit floating point value which is more suitable for vortex extraction is calculated, and the value is stored as a three-dimensional matrix.
Obtaining a weight value through a vortex characteristic detection method, if the weight value of a point is smaller than a set threshold value, the point is not the vortex point, the point smaller than the set threshold value is set to be 0, a weight matrix is pruned into a sparse matrix through threshold pruning,
the two-dimensional weight matrix pruning is as follows:
storing by using the compressed idle rows and the relative indexes, and storing the weight of the effective points contained in the matrix and the grid coordinates corresponding to the effective points in data, wherein only non-zero information in the data is reserved;
adopting four arrays val, i_index, j_index and k_index to store data;
val stores non-zero values in the valid point weights;
the i_index stores grid x coordinates corresponding to the effective point weight;
j_index stores grid y coordinates corresponding to the valid point weight;
the k_index stores grid z coordinates corresponding to the valid point weights.
When time sequence data is combined and quantized, time sequence data is firstly combined, data is read and processed, weights are then stored, and finally the weights are shared.
The data reading and processing comprises a, reading data of n files, b, respectively pruning threshold data of the n data; c. respectively adding relative indexes to n data compression idle rows for storage; d. each time data is added with a time attribute. For example, if the data belongs to data within time period 1, then the value of the time attribute is 1; e. creating a data area NodeTimeMerge with the same size as the data, and storing whether the point data is merged or not; f. traversing coordinates location (x, y, z) of each effective point in one time period in turn, and traversing each effective point coordinate in the next time period after traversing each effective point coordinate in one time period; g. judging whether the location (x, y, z) is merged or not, wherein the judging method is to see whether the value of NodeTimeMerge is 1 or not, and if so, returning to the step f; h. judging whether the location (x, y, z) appears again in a subsequent time period, if so, updating the effective attributes such as the point time attribute, the 32-bit floating point number weight, the location attribute and the like, and NodeTimeMerge=1; and i, after the step is finished, outputting all the combined data into a newly built MergeTime.csv file.
And storing the weights, sorting the weights in descending order according to the number of times of occurrence of the weights, forming a weight array, and storing the weight array.
The shared weight value is represented by respectively using serial number indexes of the weights to which the 32-bit floating point number belongs, and the original weight matrix is changed into an integer matrix and a weight array which is the shared weight value.
Compared with the prior art, the technical scheme has the beneficial effects that: 1. compared with other inventions, the invention has the innovation that the timeliness of data use is paid attention to, the timeliness of time sequence data is utilized, the redundancy of the data is further solved, the transmission speed of the data is improved, and the data storage space is saved. 2. Compared with other inventions, the invention achieves the effect of being better than other data compression. 3. The invention is suitable for data compression with time sequence.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention.
FIG. 2 is a flow chart of data reading and processing in accordance with an embodiment of the present invention.
Fig. 3 is a schematic diagram of the structure of the weight matrix converted into an integer matrix and sharing weights.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is apparent that the described embodiment is only one embodiment of the present invention, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
As can be seen from the accompanying drawings, the fluid data compression method based on the time sequence features mainly comprises the following three steps: vortex feature extraction, data pruning, time sequence data merging and quantization.
Vortex signature detection
Vortex feature extraction is an effective way to solve the problem of vortex data redundancy. For unknown data, it is not possible to find a vortex presence. The calculated fluid data may be processed by existing vortex detection algorithms. Two necessary conditions for determining the vortex: 1. the vortex must be a closed cycle; 2. the geometry of the vortex core has Galileo invariance, i.e. at all inertiasThe physical characteristics in the system are the same. In this technique, use is made ofCriteria, Q criteria, < >>The vortex characteristic extraction is realized by combining the methods of vortex quantity and the like.
criteria: the eigenvalue of the velocity tensor J is used to determine the local streamline characteristics that follow a point movement under a certain reference frame. At the center of the vortex, J should have a relatively complex eigenvalue. The eigenvalues of J, X satisfy the following equation:
wherein the method comprises the steps of
tr (J) represents the trace of matrix J, det (J) represents the determinant of matrix J,
j has complex eigenvalues when the following equation holds:
q criterion: the second component Q of the velocity gradient tensor J is a positive number and the pressure value here is smaller than the pressure value of the surrounding points. The second component Q is defined as,
Is->A centered characteristic value of the three characteristic values of (a) if a coordinate point in the fluid data corresponds to +.>If the condition is satisfied, then this point is indicated as a point in the vortex.
Vorticity: the number of rings may generally describe the speed of rotation and the size of the range of rotation, marked by the rotation of the fluid
Will beCriteria, Q criteria, < >>The method and the vortex quantity and other methods give a certain weight proportion to the obtained value, a 32-bit floating point value which is more suitable for vortex extraction is calculated, and the value is stored as a three-dimensional matrix.
Data pruning for a single time step
(1) By the vortex characteristic detection method, a weight value is obtained, and the weight value is known by the vortex detection method: if the weight value of a point is less than a certain threshold, then the point is not a point of swirl. Points less than a certain threshold are set to 0. Therefore, we prune a weight matrix into a sparse matrix by thresholding.
The two-dimensional weight matrix pruning is simply exemplified below;
(2) And storing by adding the relative index to the compressed idle row, and storing the weight of the effective point contained in the matrix and the grid coordinates corresponding to the effective point, wherein only non-zero information in the data is reserved.
Four arrays val, i_index, j_index, k_index are adopted for data storage.
Val stores non-zero values in the valid point weights;
the i_index stores grid x coordinates corresponding to the effective point weight;
j_index stores grid y coordinates corresponding to the valid point weight;
the k_index stores grid z coordinates corresponding to the valid point weights.
Time sequence data merging and quantization
(1) Time sequence data merging algorithm
Data reading and processing
a. The data of n files are read out,
b. threshold data pruning is performed on the n data respectively,
c. the n data compression idle rows are respectively stored with relative indexes,
d. each time data is added with a time attribute. For example, if the data belongs to data within time period 1, then the value of the time attribute is 1,
e. creating a data area NodeTimeMerge with the same size as the data for storing whether the point data is merged,
f. the coordinate location (x, y, z) of each valid point of one time period is traversed in turn, and after the traversing of each valid point coordinate of one time period is completed, each valid point coordinate of the next time period is traversed,
g. judging whether the location (x, y, z) is merged or not by judging whether the value of NodeTimeMerge is 1 or not, if so, returning to the step f,
h. judging whether the location (x, y, z) appears again in a subsequent time period, if so, updating the valid attributes such as the point time attribute, the 32-bit floating point number weight, the location and the like, and NodeTimeMerge=1.
i. And after the step is completed, outputting all the combined data into a newly built MergeTime.csv file.
(2) The weights are stored. And sorting the weights in descending order according to the number of times of occurrence of the weights from high to low to form a weight array, and storing the weight array.
(3) The weights are shared. The weight values expressed by the 32-bit floating point number are respectively expressed by the serial numbers of the weight values to which the 32-bit floating point number belongs, and the original weight value matrix is changed into an integer matrix and a weight value array (shared weight value) to express.
The foregoing description of the preferred embodiments of the present disclosure is provided only and not intended to limit the disclosure so that various modifications and changes may be made to the present disclosure by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.
Claims (7)
1. A method of compressing fluid data based on timing characteristics, comprising the steps of:
(1) Turbine feature detection, namely processing calculated fluid data through a vortex detection algorithm;
(2) Pruning a single time step data;
(3) Time sequential data is combined and quantized.
2. The method of time series characteristic based fluid data compression of claim 1, wherein,
in step (1), use is made ofCriteria, Q criteria, < >>The method of combining the vortex quantity method realizes vortex characteristic extraction,
criteria: the eigenvalues of the velocity tensor J are used to determine the local streamline characteristics that follow a point in a certain reference frame, at the center of the vortex, J should have a relatively complex eigenvalue, J eigenvalue, X satisfies the following equation:
tr (J) represents the trace of J; det (J) represents the determinant of J,
j has complex eigenvalues when the following equation holds:
q criterion: the second component Q of the velocity gradient tensor J is positive and the pressure value is here smaller than the pressure value of the surrounding points, the second component Q being defined as,
Is->A centered characteristic value of the three characteristic values of (a) if a coordinate point in the fluid data corresponds to +.>If the condition of (2) indicates that the point is a point in the vortex;
vorticity: the number of rings, which generally describe the speed of rotation and the size of the range of rotation, marks the rotation of the fluid,
3. The method of time series feature based fluid data compression of claim 2,
obtaining a weight value through a vortex characteristic detection method, if the weight value of a point is smaller than a set threshold value, the point is not the vortex point, the point smaller than the set threshold value is set to be 0, a weight matrix is pruned into a sparse matrix through threshold pruning,
the two-dimensional weight matrix pruning is as follows:
storing by using the compressed idle rows and the relative indexes, and storing the weight of the effective points contained in the matrix and the grid coordinates corresponding to the effective points in data, wherein only non-zero information in the data is reserved;
adopting four arrays val, i_index, j_index and k_index to store data;
val stores non-zero values in the valid point weights;
the i_index stores grid x coordinates corresponding to the effective point weight;
j_index stores grid y coordinates corresponding to the valid point weight;
the k_index stores grid z coordinates corresponding to the valid point weights.
4. The method for compressing fluid data based on time series characteristics as recited in claim 3, wherein,
when time sequence data is combined and quantized, time sequence data is firstly combined, data is read and processed, weights are then stored, and finally the weights are shared.
5. The method of time series characteristic based fluid data compression of claim 4, wherein,
the data reading and processing comprises a, reading data of n files, b, respectively pruning threshold data of the n data; c. respectively adding relative indexes to n data compression idle rows for storage; d. adding a time attribute to each piece of time data respectively, wherein if the data belongs to the data in the time period 1, the value of the time attribute is 1; e. creating a data area NodeTimeMerge with the same size as the data, and storing whether the point data is merged or not; f. traversing coordinates location (x, y, z) of each effective point in one time period in turn, and traversing each effective point coordinate in the next time period after traversing each effective point coordinate in one time period; g. judging whether the location (x, y, z) is merged or not, wherein the judging method is to see whether the value of NodeTimeMerge is 1 or not, and if so, returning to the step f; h. judging whether the location (x, y, z) appears again in a subsequent time period, if so, updating the point time attribute, the 32-bit floating point number weight, the location attribute and NodeTimeMerge=1; and i, after the step is finished, outputting all the combined data into a newly built MergeTime.csv file.
6. The method of time series characteristic based fluid data compression of claim 5,
and storing the weights, sorting the weights in descending order according to the number of times of occurrence of the weights, forming a weight array, and storing the weight array.
7. The method of time series characteristic based fluid data compression of claim 6, wherein,
the shared weight value is represented by respectively using serial number indexes of the weights to which the 32-bit floating point number belongs, and the original weight matrix is changed into an integer matrix and a weight array which is the shared weight value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310200041.4A CN116069743A (en) | 2023-03-06 | 2023-03-06 | Fluid data compression method based on time sequence characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310200041.4A CN116069743A (en) | 2023-03-06 | 2023-03-06 | Fluid data compression method based on time sequence characteristics |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116069743A true CN116069743A (en) | 2023-05-05 |
Family
ID=86171543
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310200041.4A Pending CN116069743A (en) | 2023-03-06 | 2023-03-06 | Fluid data compression method based on time sequence characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116069743A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117650791A (en) * | 2024-01-30 | 2024-03-05 | 苏芯物联技术(南京)有限公司 | Welding history airflow data compression method integrating welding process mechanism |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160377504A1 (en) * | 2015-06-29 | 2016-12-29 | Rolls-Royce Plc | Vortex identification methods and tools |
CN108153483A (en) * | 2016-12-06 | 2018-06-12 | 南京南瑞继保电气有限公司 | A kind of time series data compression method based on attribute grouping |
CN109871362A (en) * | 2019-02-13 | 2019-06-11 | 北京航空航天大学 | A kind of data compression method towards streaming time series data |
CN111652366A (en) * | 2020-05-09 | 2020-09-11 | 哈尔滨工业大学 | Combined neural network model compression method based on channel pruning and quantitative training |
CN112668256A (en) * | 2021-01-14 | 2021-04-16 | 昆明理工大学 | Vertical seam type fishway vortex identification method |
CN114037739A (en) * | 2021-10-29 | 2022-02-11 | 齐鲁工业大学 | Flow field feature extraction method and device based on machine learning and storage medium |
-
2023
- 2023-03-06 CN CN202310200041.4A patent/CN116069743A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160377504A1 (en) * | 2015-06-29 | 2016-12-29 | Rolls-Royce Plc | Vortex identification methods and tools |
CN108153483A (en) * | 2016-12-06 | 2018-06-12 | 南京南瑞继保电气有限公司 | A kind of time series data compression method based on attribute grouping |
CN109871362A (en) * | 2019-02-13 | 2019-06-11 | 北京航空航天大学 | A kind of data compression method towards streaming time series data |
CN111652366A (en) * | 2020-05-09 | 2020-09-11 | 哈尔滨工业大学 | Combined neural network model compression method based on channel pruning and quantitative training |
CN112668256A (en) * | 2021-01-14 | 2021-04-16 | 昆明理工大学 | Vertical seam type fishway vortex identification method |
CN114037739A (en) * | 2021-10-29 | 2022-02-11 | 齐鲁工业大学 | Flow field feature extraction method and device based on machine learning and storage medium |
Non-Patent Citations (2)
Title |
---|
CORNEL CONSTANTINESCU: "Fast and Efficient Compression of Next Generation Sequencing Data", 《2018 DATA COMPRESSION CONFERENCE》 * |
王义乾: "第三代涡识别方法及其应用综述", 《水动力学研究与进展(A辑)》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117650791A (en) * | 2024-01-30 | 2024-03-05 | 苏芯物联技术(南京)有限公司 | Welding history airflow data compression method integrating welding process mechanism |
CN117650791B (en) * | 2024-01-30 | 2024-04-05 | 苏芯物联技术(南京)有限公司 | Welding history airflow data compression method integrating welding process mechanism |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | O-cnn: Octree-based convolutional neural networks for 3d shape analysis | |
Esteves et al. | Learning so (3) equivariant representations with spherical cnns | |
Zhi et al. | LightNet: A Lightweight 3D Convolutional Neural Network for Real-Time 3D Object Recognition. | |
Ma et al. | Binary volumetric convolutional neural networks for 3-D object recognition | |
CN111382300B (en) | Multi-view three-dimensional model retrieval method and system based on pairing depth feature learning | |
Zhu et al. | Adafit: Rethinking learning-based normal estimation on point clouds | |
CN102521843B (en) | Three-dimensional human body motion analysis and synthesis method based on manifold learning | |
Liu et al. | Optimization-based key frame extraction for motion capture animation | |
CN116069743A (en) | Fluid data compression method based on time sequence characteristics | |
CN109902714A (en) | A kind of multi-modality medical image search method based on more figure regularization depth Hash | |
Zhao et al. | Michelangelo: Conditional 3d shape generation based on shape-image-text aligned latent representation | |
Hou et al. | Hitpr: Hierarchical transformer for place recognition in point cloud | |
Wang et al. | 3D point cloud analysis and classification in large-scale scene based on deep learning | |
CN104268285A (en) | Motion capture database searching method based on dynamic bending algorithm | |
Zhao et al. | MobileNet-Yolo based wildlife detection model: A case study in Yunnan Tongbiguan Nature Reserve, China | |
Liu et al. | A lightweight object detection algorithm for remote sensing images based on attention mechanism and YOLOv5s | |
Cao et al. | Detection of microalgae objects based on the Improved YOLOv3 model | |
CN110851627A (en) | Method for describing sun black subgroup in full-sun image | |
CN112580645B (en) | Unet semantic segmentation method based on convolution sparse coding | |
Han et al. | QuickFPS: Architecture and algorithm co-design for farthest point sampling in large-scale point clouds | |
Li et al. | Towards communication-efficient digital twin via ai-powered transmission and reconstruction | |
CN113159117A (en) | Streamline generation method and device | |
Huang et al. | Adaptive-Attention Completing Network for Remote Sensing Image | |
CN104272324A (en) | Method and apparatus for generating shape descriptor of a model | |
CN116664874B (en) | Single-stage fine-granularity light-weight point cloud 3D target detection system and method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20230505 |
|
RJ01 | Rejection of invention patent application after publication |