CN112733067A - Data set selection method for robot target detection algorithm - Google Patents
Data set selection method for robot target detection algorithm Download PDFInfo
- Publication number
- CN112733067A CN112733067A CN202011542396.4A CN202011542396A CN112733067A CN 112733067 A CN112733067 A CN 112733067A CN 202011542396 A CN202011542396 A CN 202011542396A CN 112733067 A CN112733067 A CN 112733067A
- Authority
- CN
- China
- Prior art keywords
- data set
- similarity
- target detection
- row vector
- robot
- 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.)
- Granted
Links
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Image Analysis (AREA)
- Manipulator (AREA)
Abstract
The invention provides a data set selection method for a robot target detection algorithm, aiming at the problem of how to select a data set by a robot target detection algorithm model in various application scenes. According to the method, the machine learning method is used, the data set can be automatically selected for training and testing the model according to different requirements of the algorithm model, an artificial experience method can be effectively replaced, and meanwhile the robustness and generalization performance of the algorithm model are improved. And the method provided by the invention extracts the metadata characteristics of the data set influencing the detection effect of the model according to the test conclusion, re-encodes the row vectors and can effectively reduce the similarity value of the next matching. Through continuous iterative update learning, row vector coding of the data set in the step one is improved, a more appropriate data set can be provided for a next new robot target recognition algorithm model, and the robust performance and generalization performance of the algorithm model are improved.
Description
Technical Field
The invention relates to the field of machine learning, in particular to a data set selection method facing a robot target detection algorithm.
Background
With the rapid development of artificial intelligence deep learning, computer vision-based target detection techniques have been applied to various scenes. In particular, in the field of robots, target detection technologies for industrial robots, service robots, unmanned aerial vehicles, security monitoring and other scenes have been developed increasingly. Different application scenarios typically select corresponding data sets for training and testing of the model. In order to select a suitable data set for an algorithm model developer so that the algorithm model can achieve the optimal performance, the method becomes a research hotspot of each algorithm model developer in recent years.
The existing robot target detection algorithm model data set recommendation research generally adopts a method based on manual experience, and the selection with human subjectivity usually needs to be subjected to a large number of model parameter adjusting processes subsequently.
Disclosure of Invention
The purpose of the invention is: a data set selection method based on machine learning is provided for a robot target detection algorithm model.
In order to achieve the above object, the technical solution of the present invention is to provide a data set selection method for a robot target detection algorithm, which is characterized by comprising the following steps:
step 1: performing row vector coding on the existing metadata characteristics of each type of data set, wherein the method comprises the following steps:
step 101: each metadata feature contains different numbers of feature elements, and if the total number of all the feature elements of all the metadata features is n, a 1 x n-dimensional matrix is constructed, wherein each element in the matrix corresponds to one feature element in one metadata feature;
step 102: setting all elements of the 1 xn dimensional matrix obtained in step 101 to 0, a set of row vectors with n 0 s is obtained for each type of data set
Step 103: obtaining the row vector corresponding to each kind of data set
The method comprises the following steps:
if the current data set contains a certain feature element in a certain metadata feature, the corresponding row vector obtained in step 102 is used
The value of the corresponding element in (1) is set from 0, so that the current data set corresponds to a 1 xn dimensional matrix containing a plurality of element values of 1, and the 1 xn dimensional matrix is defined as a row vector
Step 2: determining a target detection object required by a robot target detection algorithm model, and performing row vector coding on the metadata characteristics of the target detection object by using the same method as the step 1 so as to establish a row vector corresponding to the metadata characteristics of the target detection object
And step 3: based on the row vector obtained in step 1
And the row vector obtained in step 2
Calculating the similarity I between the metadata features of a target detection object related to the robot field and the metadata features of an existing data set, wherein the higher the similarity I is, the more matched the current data set and the target detection object is, and the lower the similarity is, the more unmatched the current data set and the target detection object is;
and 4, step 4: taking the data set with the highest similarity in the step 3 as a reference data set recommended by a target detection object in the robot field, respectively carrying out similarity calculation on the rest data sets and the reference data set, and calculating a similarity value II by using a row vector corresponding to each data set;
and 5: and (4) giving a similarity threshold value I and a similarity threshold value II, and taking all data sets with the similarity value I higher than the similarity threshold value I and the similarity value II higher than the similarity threshold value II and the reference data set determined in the step 4 as recommended data sets of the target detection object.
Preferably, the row vector of the current data set is calculated by using a formula of cosine similarity
The row vector corresponding to the target detection object
The calculated distance is used as a value of similarity one between the target detection object and the current data set; a larger value of the cosine similarity indicates a better match between the target detection object and the current data set.
Preferably, the specific calculation method of the cosine similarity is as follows:
cos (A, B) represents a row vector
And the row vector
Cosine similarity of (1), where A represents a row vector
Modulo of (a), B represents a row vector
The die of (1).
Preferably, in step 3, the calculated similarity values of all the data sets are sorted in descending order from high to low, and then the descending order of all the corresponding data sets is completed.
Preferably, in step 4, all the values of the similarity two are arranged in descending order from high to low.
Preferably, after the step 5, the method further comprises the following steps:
step 6: after the recommended data set in the robot field obtained in the step 5 is used for training and testing, the metadata characteristics of the data set influencing the detection effect of the model are extracted according to the test conclusion, and the row vectors are recoded, so that the similarity value of the next matching can be effectively reduced; through continuous iterative update learning, row vector coding of the data set in the step 1 is improved, a more appropriate data set is provided for a next new robot target recognition algorithm model, and the robust performance and the generalization performance of the algorithm model are improved.
The invention provides a method for recommending a proper data set for a robot target detection algorithm model developer, aiming at the problem of how to select the data set by a robot target detection algorithm model in various application scenes. According to the method, the machine learning method is used, the data set can be automatically selected for training and testing the model according to different requirements of the algorithm model, an artificial experience method can be effectively replaced, and meanwhile the robustness and generalization performance of the algorithm model are improved. The method provided by the invention analyzes the metadata characteristics of the data set which influence the detection effect of the model, and recodes the row vector, thereby effectively reducing the similarity value of the next matching. Through continuous iterative updating, a more appropriate data set can be provided for the next robot target recognition algorithm model, and the robust performance and the generalization performance of the algorithm model are improved.
Drawings
FIG. 1 is a schematic overall flow chart of a data set selection method for a robot target detection algorithm model according to the present invention;
FIG. 2 is a diagram illustrating row vector encoding of metadata features according to the present invention.
Detailed Description
The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
As shown in fig. 1, the method for selecting a data set for a robot-oriented target recognition algorithm provided by the present invention includes the following steps:
the method comprises the following steps: and carrying out row vector coding on the existing metadata characteristics of each type of data set.
The metadata characteristics include an application scene, a target detection object category, a target detection object size, illumination brightness, and the like. The method specifically comprises the following steps that an algorithm model developer target detection object and each metadata feature traverse to establish a similarity matching relation:
each metadata feature contains different numbers of feature elements, for example, a robot application scene is a metadata feature, and the metadata feature contains a plurality of feature elements corresponding to different scenes such as home, shopping mall, garden, and the like. Assuming that the total number of all feature elements of all metadata features is n, a 1 × n dimensional matrix is constructed. When initializing, all elements of the 1 × n dimensional matrix are set to 0, so that a group of row vectors with n 0 s can be obtained
In n characteristic elements of the 1 Xn dimensional matrix, the first metadata characteristic is a robot application scene, 1 st to nth1The characteristic elements belong to a first metadata characteristic and respectively correspond to different scenes such as home, shopping malls, parks and the like. N th1+1 characteristic elements to the nth2The feature element belongs to a second metadata feature, the second metadata feature being a target detection object class. N th2+1 to nth feature elements belonging to the third elementAnd the data characteristic and the third metadata characteristic are target detection object sizes. And coding in the initialized 1 × n dimensional matrix according to the corresponding relation. During encoding, if a current data set contains a certain feature element in a certain metadata feature, the value of the corresponding element in the 1 xn dimensional matrix is set from 0 to 1, so that each data set corresponds to a 1 xn dimensional matrix containing a plurality of element values of 1, and the 1 xn dimensional matrix is defined as a row vector
Each data set can therefore be represented as a different row vector
Step two: and (4) determining a target detection object required by the robot target detection algorithm model, and performing row vector coding on the metadata characteristics of the target detection object by using the method same as the step one. Similar to the first step, if the target detection object contains a certain feature element in a certain metadata feature, the value of the corresponding element in the 1 × n dimensional matrix is set from 0 to 1, so as to establish a row vector corresponding to the metadata feature of the target detection object
Step three: and calculating the similarity I between the metadata features of the target detection object related to the robot field and the metadata features of the existing data set, wherein the higher the similarity I is, the more matched the current data set and the target detection object is, and the lower the similarity I is, the more unmatched the current data set and the target detection object is.
Calculating a row vector of a current data set by using a formula of cosine similarity
Line vector corresponding to target detection object
Will count the distance betweenThe calculated distance is used as a value of the similarity of the target detection object and the current data set. A larger value of the cosine similarity indicates a better match between the target detection object and the current data set. The specific calculation method of the cosine similarity is shown as the following formula:
cos (A, B) represents a row vector
And the row vector
Cosine similarity of (1), where A represents a row vector
Modulo of (a), B represents a row vector
The die of (1).
And sequencing the similarity values of all the calculated data sets in a descending order from high to low, and further finishing the descending order of all the corresponding data sets.
Step four: and taking the data set ranked at the first in the third step as a reference data set recommended by the target detection object in the robot field, and respectively carrying out similarity calculation on the rest data sets and the reference data set. Similar to the first step, the second step and the third step, the values of the second similarity obtained by calculating the row vector corresponding to each data set are arranged in descending order from high to low.
Step five: and giving a similarity threshold value I and a similarity threshold value II, and taking all data sets of which the similarity value I is higher than the similarity threshold value I and the similarity value II is higher than the similarity threshold value II and the reference data set determined in the step four as recommended data sets of the target detection object.
Step six: and after the recommended data set in the robot field obtained in the step five is used for training and testing, extracting data set metadata characteristics influencing the detection effect of the model according to the test conclusion, and recoding the row vectors, so that the similarity value of low-time matching can be effectively reduced. Through continuous iterative update learning, row vector coding of the data set in the step one is improved, a more appropriate data set can be provided for a next new robot target recognition algorithm model, and the robust performance and generalization performance of the algorithm model are improved.
Claims (6)
1. A data set selection method for a robot target detection algorithm is characterized by comprising the following steps:
step 1: performing row vector coding on the existing metadata characteristics of each type of data set, wherein the method comprises the following steps:
step 101: each metadata feature contains different numbers of feature elements, and if the total number of all the feature elements of all the metadata features is n, a 1 x n-dimensional matrix is constructed, wherein each element in the matrix corresponds to one feature element in one metadata feature;
step 102: setting all elements of the 1 xn dimensional matrix obtained in step 101 to 0, a set of row vectors with n 0 s is obtained for each type of data set
Step 103: obtaining the row vector corresponding to each kind of data set
The method comprises the following steps:
if the current data set contains a certain feature element in a certain metadata feature, the corresponding row vector obtained in step 102 is used
The value of the corresponding element in (1) is set from 0, so that the current data set corresponds to a 1 xn dimensional matrix containing a plurality of element values of 1, and the 1 xn dimensional matrix is defined as a row vector
Step 2: determining a target detection object required by a robot target detection algorithm model, and performing row vector coding on the metadata characteristics of the target detection object by using the same method as the step 1 so as to establish a row vector corresponding to the metadata characteristics of the target detection object
And step 3: based on the row vector obtained in step 1
And the row vector obtained in step 2
Calculating the similarity I between the metadata features of a target detection object related to the robot field and the metadata features of an existing data set, wherein the higher the similarity I is, the more matched the current data set and the target detection object is, and the lower the similarity is, the more unmatched the current data set and the target detection object is;
and 4, step 4: taking the data set with the highest similarity in the step 3 as a reference data set recommended by a target detection object in the robot field, respectively carrying out similarity calculation on the rest data sets and the reference data set, and calculating a similarity value II by using a row vector corresponding to each data set;
and 5: and (4) giving a similarity threshold value I and a similarity threshold value II, and taking all data sets with the similarity value I higher than the similarity threshold value I and the similarity value II higher than the similarity threshold value II and the reference data set determined in the step 4 as recommended data sets of the target detection object.
2. The method of claim 1, wherein the row vector of the current dataset is calculated using a cosine similarity formula
The row vector corresponding to the target detection object
The calculated distance is used as a value of similarity one between the target detection object and the current data set; a larger value of the cosine similarity indicates a better match between the target detection object and the current data set.
3. The method for selecting a data set for a robot-oriented object detection algorithm according to claim 2, wherein the specific calculation method of the cosine similarity is as follows:
cos (A, B) represents a row vector
And the row vector
Cosine similarity of (1), where A represents a row vector
Modulo of (a), B represents a row vector
The die of (1).
4. The method for selecting a data set oriented to a robot target detection algorithm according to claim 1, wherein in step 3, the calculated similarity values of all the data sets are sorted in descending order from high to low, and then the sorting of all the corresponding data sets is completed.
5. The method for selecting a data set for a robot-oriented object detection algorithm according to claim 1, wherein in step 4, all the similarity two values are arranged in descending order from high to low.
6. A method for selecting a data set for a robot-oriented object detection algorithm as claimed in claim 1, further comprising after said step 5:
step 6: after the recommended data set in the robot field obtained in the step 5 is used for training and testing, the metadata characteristics of the data set influencing the detection effect of the model are extracted according to the test conclusion, and the row vectors are recoded, so that the similarity value of the next matching can be effectively reduced; through continuous iterative update learning, row vector coding of the data set in the step 1 is improved, a more appropriate data set is provided for a next new robot target recognition algorithm model, and the robust performance and the generalization performance of the algorithm model are improved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011542396.4A CN112733067B (en) | 2020-12-22 | 2020-12-22 | Data set selection method for robot target detection algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011542396.4A CN112733067B (en) | 2020-12-22 | 2020-12-22 | Data set selection method for robot target detection algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112733067A true CN112733067A (en) | 2021-04-30 |
| CN112733067B CN112733067B (en) | 2023-05-09 |
Family
ID=75604750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011542396.4A Active CN112733067B (en) | 2020-12-22 | 2020-12-22 | Data set selection method for robot target detection algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112733067B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114241305A (en) * | 2021-12-07 | 2022-03-25 | 中原动力智能机器人有限公司 | Target detection model updating method and device |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107657216A (en) * | 2017-09-11 | 2018-02-02 | 安徽慧视金瞳科技有限公司 | 1 to the 1 face feature vector comparison method based on interference characteristic vector data collection |
| CN107749034A (en) * | 2017-11-17 | 2018-03-02 | 浙江工业大学 | A kind of safe friend recommendation method in social networks |
| CN108491872A (en) * | 2018-03-16 | 2018-09-04 | 深圳市商汤科技有限公司 | Target recognition methods and device, electronic equipment, program and storage medium again |
| CN108509467A (en) * | 2017-05-04 | 2018-09-07 | 宁波数联软件有限公司 | A kind of sea-freight quotation commending system and method based on User action log |
| CN109145111A (en) * | 2018-07-27 | 2019-01-04 | 深圳市翼海云峰科技有限公司 | A kind of multiple features text data similarity calculating method based on machine learning |
| CN109255586A (en) * | 2018-08-24 | 2019-01-22 | 安徽讯飞智能科技有限公司 | A kind of online personalized recommendation method that E-Governance Oriented is handled affairs |
| CN109615466A (en) * | 2018-11-27 | 2019-04-12 | 浙江工商大学 | The mixed method of commending contents and collaborative filtering recommending towards mobile ordering system |
| CN110738245A (en) * | 2019-09-29 | 2020-01-31 | 上海大学 | automatic clustering algorithm selection system and method for scientific data analysis |
| CN110990383A (en) * | 2019-10-14 | 2020-04-10 | 同济大学 | Similarity calculation method based on industrial big data set |
| CN111460316A (en) * | 2020-03-20 | 2020-07-28 | 南京邮电大学 | Knowledge system-oriented personalized recommendation method and computer storage medium |
| CN111553193A (en) * | 2020-04-01 | 2020-08-18 | 东南大学 | Visual SLAM closed-loop detection method based on lightweight deep neural network |
| CN111834011A (en) * | 2020-07-10 | 2020-10-27 | 华东师范大学 | Long-term care-for-the-elderly oriented collaborative interactive service recommendation method |
-
2020
- 2020-12-22 CN CN202011542396.4A patent/CN112733067B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108509467A (en) * | 2017-05-04 | 2018-09-07 | 宁波数联软件有限公司 | A kind of sea-freight quotation commending system and method based on User action log |
| CN107657216A (en) * | 2017-09-11 | 2018-02-02 | 安徽慧视金瞳科技有限公司 | 1 to the 1 face feature vector comparison method based on interference characteristic vector data collection |
| CN107749034A (en) * | 2017-11-17 | 2018-03-02 | 浙江工业大学 | A kind of safe friend recommendation method in social networks |
| CN108491872A (en) * | 2018-03-16 | 2018-09-04 | 深圳市商汤科技有限公司 | Target recognition methods and device, electronic equipment, program and storage medium again |
| CN109145111A (en) * | 2018-07-27 | 2019-01-04 | 深圳市翼海云峰科技有限公司 | A kind of multiple features text data similarity calculating method based on machine learning |
| CN109255586A (en) * | 2018-08-24 | 2019-01-22 | 安徽讯飞智能科技有限公司 | A kind of online personalized recommendation method that E-Governance Oriented is handled affairs |
| CN109615466A (en) * | 2018-11-27 | 2019-04-12 | 浙江工商大学 | The mixed method of commending contents and collaborative filtering recommending towards mobile ordering system |
| CN110738245A (en) * | 2019-09-29 | 2020-01-31 | 上海大学 | automatic clustering algorithm selection system and method for scientific data analysis |
| CN110990383A (en) * | 2019-10-14 | 2020-04-10 | 同济大学 | Similarity calculation method based on industrial big data set |
| CN111460316A (en) * | 2020-03-20 | 2020-07-28 | 南京邮电大学 | Knowledge system-oriented personalized recommendation method and computer storage medium |
| CN111553193A (en) * | 2020-04-01 | 2020-08-18 | 东南大学 | Visual SLAM closed-loop detection method based on lightweight deep neural network |
| CN111834011A (en) * | 2020-07-10 | 2020-10-27 | 华东师范大学 | Long-term care-for-the-elderly oriented collaborative interactive service recommendation method |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114241305A (en) * | 2021-12-07 | 2022-03-25 | 中原动力智能机器人有限公司 | Target detection model updating method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112733067B (en) | 2023-05-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20240046094A1 (en) | Semi-Supervised Person Re-Identification Using Multi-View Clustering | |
| CN109800648B (en) | Face detection and recognition method and device based on face key point correction | |
| CN105138973B (en) | The method and apparatus of face authentication | |
| CN109815801A (en) | Face identification method and device based on deep learning | |
| CN109063649B (en) | Pedestrian re-identification method based on twin pedestrian alignment residual error network | |
| CN110414550B (en) | Training method, device and system of face recognition model and computer readable medium | |
| CN111414849A (en) | Face recognition method based on evolution convolutional neural network | |
| CN106780639B (en) | Hash coding method based on significance characteristic sparse embedding and extreme learning machine | |
| CN110942091A (en) | Semi-supervised few-sample image classification method for searching reliable abnormal data center | |
| CN110175615A (en) | The adaptive visual position recognition methods in model training method, domain and device | |
| CN114092747A (en) | Small sample image classification method based on depth element metric model mutual learning | |
| CN113850281A (en) | Data processing method and device based on MEANSHIFT optimization | |
| WO2023124278A1 (en) | Image processing model training method and apparatus, and image classification method and apparatus | |
| CN113705596A (en) | Image recognition method and device, computer equipment and storage medium | |
| Halder et al. | An evolutionary dynamic clustering based colour image segmentation | |
| CN115424050A (en) | Method and system for detecting and positioning ceramic tile surface defects and storage medium | |
| CN109886206B (en) | Three-dimensional object identification method and equipment | |
| CN110188864B (en) | Small sample learning method based on distribution representation and distribution measurement | |
| CN116844004A (en) | Point cloud automatic semantic modeling method for digital twin scene | |
| CN111079930A (en) | Method and device for determining quality parameters of data set and electronic equipment | |
| CN114781779A (en) | Unsupervised energy consumption abnormity detection method and device and storage medium | |
| CN112733067A (en) | Data set selection method for robot target detection algorithm | |
| CN111783688B (en) | Remote sensing image scene classification method based on convolutional neural network | |
| CN113743443A (en) | Image evidence classification and identification method and device | |
| CN114067155B (en) | Image classification method, device, product and storage medium based on meta learning |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |