CN113420017A - Block chain application method for acquiring robot navigation algorithm training data set - Google Patents
Block chain application method for acquiring robot navigation algorithm training data set Download PDFInfo
- Publication number
- CN113420017A CN113420017A CN202110685237.8A CN202110685237A CN113420017A CN 113420017 A CN113420017 A CN 113420017A CN 202110685237 A CN202110685237 A CN 202110685237A CN 113420017 A CN113420017 A CN 113420017A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- database
- local application
- account book
- 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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses a block chain application method for acquiring a robot navigation algorithm training data set, which adopts the technical scheme that the method comprises the following steps of: building a block chain network, wherein the block chain network comprises a plurality of public application nodes, and the public application nodes are used for storing data uploaded by a user in real time; a data acquisition step: constructing a local application node, acquiring data conforming to the data type from the public application node at regular time according to the preset data type, and storing the data to the local application node; a data training step: and training the data of the local application nodes by using a semi-supervised learning algorithm to obtain a machine vision model. By the method, diversified data sets are easy to obtain, so that the trained model can adapt to complex scenes.
Description
Technical Field
The invention relates to the field of machine vision, in particular to a block chain application method for acquiring a robot navigation algorithm training data set.
Background
Machine vision is an important branch of the field of artificial intelligence in recent years. The vision-based navigation uses a computer vision algorithm and an optical sensor, comprises a laser-based range finder and a photometric camera using a CCD array, converts an image signal into a digital signal, and guides a robot to make corresponding actions through a series of processing such as image preprocessing, feature extraction, machine learning and the like. Machine learning is the key of whole navigation, the data sets trained at present are from laboratories or individual large-scale science and technology companies, the data sets are too single, and the trained algorithm cannot adapt to various complex scenes such as weather, light, colors and the like.
In summary, training a machine learning algorithm using a traditional centralized data set sample has the following problems: 1. the data acquisition is difficult: machine learning sample diversity largely determines the performance of final machine recognition, but large-scale data acquisition is quite difficult for a typical individual or business, limiting the development of machine vision to some extent. 2. Data singulation: data are only mastered in a few large Internet companies, the data are only some user behavior data and are not diversified, and therefore the trained objective function cannot adapt to the conditions under various complex scenes, and the navigation visual precision is greatly reduced.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a block chain application method for acquiring a robot navigation algorithm training data set, and diversified data sets are easily acquired by the method, so that a model obtained by training can adapt to a complex scene.
In order to achieve the purpose, the invention provides the following technical scheme: a block chain application method for acquiring a robot navigation algorithm training data set comprises the following steps:
building a block network: building a block chain network, wherein the block chain network comprises a plurality of public application nodes, and the public application nodes are used for storing data uploaded by a user in real time;
a data acquisition step: constructing a local application node, acquiring data conforming to a data type from the public application node at regular time according to a preset data type, and storing the data to the local application node;
a data training step: and training the data of the local application nodes by using a semi-supervised learning algorithm to obtain a machine vision model.
As a further improvement of the present invention, the local application node is configured with a local application database, and the data acquisition step includes a data uplink sub-step and a data screening sub-step;
the data uplink sub-step comprises the steps that the terminal equipment responds to a user instruction to format data into a json format, sets the data into a preset data type, packs the data and uploads the packed data to any public application node;
the data warehousing substep comprises the steps of obtaining the maximum height of the current account book of the database as Max, setting the height of the current obtained account book as Max +1, obtaining data from the corresponding account book of the public application node at the current obtained account book height of Max +1 every preset obtaining interval time, screening the data, storing the screened data in the local application database if the data is obtained after screening, and adding 1 to the current account book height.
As a further improvement of the present invention, the data screening in the data warehousing substep specifically comprises: judging whether the corresponding account book is empty, and if the account book is empty, adding 1 to the height of the current account book; if not, reading the data in the corresponding account book; comparing the type of the read data with a preset data type, and if the type of the read data is the preset data type, storing the read data in the local application database; and if the type of the read data is different from the preset data type, not storing the read data.
As a further improvement of the present invention, the local application node is further configured with an exception database, where exception data is stored in the exception database, where the exception data is data that is obtained from the local application database and cannot be used for training and has no training value, and the data training step further includes a data selection substep;
the data selection substep comprises comparing data obtained from the local application database with data obtained from the anomaly database, and selecting data obtained from the local application database and not present in the anomaly database for training.
As a further improvement of the present invention, the data training step further includes an anomaly feedback sub-step, and the anomaly feedback sub-step includes storing data which cannot be trained and has no training value in the acquired data to the anomaly database when the acquired data are trained one by one.
As a further improvement of the present invention, the acquisition interval time coincides with a time interval between the previous and subsequent block generation.
As a further improvement of the present invention, in the data uplink substep, data is uploaded to the public application node by an HTTP protocol, and in the data warehousing substep, data is acquired from a corresponding account book of the public application node by using an HTTP Restful method.
As a further improvement of the present invention, the preset data type is that the transaction type of the data is set as: and M.
As a further refinement of the invention, the data training step comprises training the data under the TensorFlow framework.
As a further improvement of the present invention, the local application database is a non-relational database.
The invention has the beneficial effects that: through the block network building step and the data acquisition step, a company needing training data can continuously acquire the training data which is continuously uploaded to the public application node by a user with the training data through the local application node, and the data is acquired according to the preset data type, so that most useless data can be screened out, and the data availability is improved. Through the data training step, the company trains the machine navigation vision through the acquired data to obtain a machine vision model. According to the characteristics of distribution, decentralization, traceability and tamper resistance of the block chain, anyone can share data on the chain, so that any user can upload experimental data to the public application nodes, the threshold for uploading the experimental data by the user is low, and each public application node can store a large amount of various experimental data. After the companies with data demands obtain the various experimental data, the navigation visual precision of the machine visual model obtained by training can be improved, and the robustness of the machine learning model is improved. The method is convenient for companies needing machine vision training to obtain diversified data sets, so that the trained model can adapt to complex scenes.
Drawings
FIG. 1 is a schematic flow chart of example 1;
FIG. 2 is a schematic flow chart of example 2.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. In which like parts are designated by like reference numerals.
Example 1:
referring to fig. 1, the method for applying a blockchain for acquiring a robot navigation algorithm training data set according to the embodiment includes a blocknetwork building step, a data acquiring step, and a data training step.
Building a block network: and building a block chain network, wherein the block chain network comprises a plurality of public application nodes, and the public application nodes are used for storing data uploaded by a user in real time.
The block chain network is composed of a plurality of nodes, according to a Byzantine fault-tolerant mechanism, the total number of verification nodes is generally more than or equal to 3f +1(f is the number of abnormal nodes), and if the number of the fault-tolerant nodes is 1, the verification nodes need to be at least four. The verification node participates in the block consensus and is the core of the whole block chain network. In order to acquire a data set shared on a block chain, a plurality of application nodes need to be configured, the application nodes do not participate in consensus, and the application nodes participate in data broadcasting and synchronization.
A data acquisition step: and constructing a local application node, and configuring a local application database in the local application node, wherein the local application database is configured to be a non-relational database. The data acquisition step includes a data uplink sub-step and a data screening sub-step.
The data uplink sub-step comprises the steps that the terminal equipment responds to a user instruction to format data into a json format, sets the data into a preset data type, packages the data and uploads the packaged data to any public application node through an HTTP (hyper text transport protocol), wherein the preset data type is set as a transaction type of the data: and M. The terminal device is a terminal capable of uploading data to the public application node, and may be a mobile terminal such as a mobile phone, a notebook, a tablet computer, or a non-mobile terminal such as a desktop computer or a server.
And the data warehousing substep comprises the steps of acquiring the maximum height of the current account book of the database as Max, setting the height of the current acquired account book as Max +1, and acquiring data from the corresponding account book of the common application node by adopting an HTTP Restful mode at intervals of preset acquisition intervals and the current height of the acquired account book Max + 1. In the data acquisition process, whether the corresponding account book is empty is judged, if the account book is empty, the height of the current account book is increased by 1, and the data acquisition from the public application node is finished. If not, reading the data in the corresponding account book; comparing the type of the read data with a preset data type, namely judging whether the transaction type of the data is as follows: m, if the transaction type of the read data is: and M, storing the read data into a local application database, and adding 1 to the current book height. If the transaction type of the read data is not: and M, not storing the read data, and adding 1 to the current book height. Wherein, the acquisition interval time is set as the time interval of the generation of the front and rear blocks.
Because accounts in the block chain are connected in order according to the block Hash in a one-way mode, training data exist in each account of each bulletin application node, the data types of the training data are various, and if the data are directly acquired from the chain to serve as the training data for machine learning, a large amount of other data which are not used for machine learning training can be obtained. In order to improve the availability of the obtained data, the transaction type of the data is limited, only the data which is in accordance with the transaction type can be adopted, and the data which is not in accordance with the transaction type is abandoned, so that the screening of the ledger data in the common application node is completed, and the batch of valuable training data is obtained.
A data training step: and training data of the local application node under a TensorFlow framework by using a semi-supervised learning algorithm to obtain a machine vision model, and guiding the robot to navigate according to the machine vision model to make accurate judgment.
Example 2:
referring to fig. 1, the difference between this embodiment and embodiment 1 is that an exception database is further configured in the local application node, and exception data is stored in the exception database, where the exception data is data that is not used for training and has no training value and is acquired from the local application database. The abnormal data is obtained in the following mode: the data training step also comprises an anomaly feedback substep, wherein the anomaly feedback substep comprises the step of storing data which cannot be trained and has no training value into an anomaly database when the obtained data are trained one by one. That is, when training is performed on data, if the data is found to be incomplete or obviously wrong, the data is indicated to have no training value and should not or cannot be used for machine vision training, and the data which has no training value and cannot be trained is stored in the abnormal database.
The data training step also comprises a data selecting sub-step; the data selection substep comprises comparing data obtained from the local application database with data obtained from the exception database, and selecting data obtained from the local application database and not present in the exception database for training.
Since the abnormal data is originally obtained from the local application database, the data of the abnormal data and the data of the local application database are consistent. When data is acquired, the data in the local application database is obtained first, and then the data in the abnormal database is discarded, so that the left data has higher availability. When a plurality of companies or a plurality of training bases of one company obtain training data from the local application node, the availability of the data obtained by the company or the subsequent training base which subsequently extracts the data can be improved through the setting of the abnormal database, the interference of the data without training value on training can be reduced, and the training time can be saved.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may occur to those skilled in the art without departing from the principle of the invention, and are considered to be within the scope of the invention.
Claims (10)
1. A block chain application method for acquiring a robot navigation algorithm training data set is characterized by comprising the following steps: the method comprises the following steps:
building a block network: building a block chain network, wherein the block chain network comprises a plurality of public application nodes, and the public application nodes are used for storing data uploaded by a user in real time;
a data acquisition step: constructing a local application node, acquiring data conforming to a data type from the public application node at regular time according to a preset data type, and storing the data to the local application node;
a data training step: and training the data of the local application nodes by using a semi-supervised learning algorithm to obtain a machine vision model.
2. The method of claim 1, wherein the method comprises: the local application node is configured with a local application database, and the data acquisition step comprises a data uplink sub-step and a data screening sub-step;
the data uplink sub-step comprises the steps that the terminal equipment responds to a user instruction to format data into a json format, sets the data into a preset data type, packs the data and uploads the packed data to any public application node;
the data warehousing substep comprises the steps of obtaining the maximum height of the current account book of the database as Max, setting the height of the current obtained account book as Max +1, obtaining data from the corresponding account book of the public application node at the current obtained account book height of Max +1 every preset obtaining interval time, screening the data, storing the screened data in the local application database if the data is obtained after screening, and adding 1 to the current account book height.
3. The blockchain application method for the acquisition of the training data set of the robot navigation algorithm according to claim 2, wherein the blockchain application method comprises the following steps: the data screening in the data warehousing substep specifically comprises the following steps: judging whether the corresponding account book is empty, and if the account book is empty, adding 1 to the height of the current account book; if not, reading the data in the corresponding account book; comparing the type of the read data with a preset data type, and if the type of the read data is the preset data type, storing the read data in the local application database; and if the type of the read data is different from the preset data type, not storing the read data.
4. The blockchain application method for the acquisition of the training data set of the robot navigation algorithm according to claim 3, wherein the blockchain application method comprises the following steps: the local application node is also provided with an abnormal database, abnormal data are stored in the abnormal database, the abnormal data are data which are obtained from the local application database and cannot be used for training and have no training value, and the data training step further comprises a data selecting sub-step;
the data selection substep comprises comparing data obtained from the local application database with data obtained from the anomaly database, and selecting data obtained from the local application database and not present in the anomaly database for training.
5. The blockchain application method for the acquisition of the training data set of the robot navigation algorithm according to claim 4, wherein the blockchain application method comprises the following steps: the data training step further comprises an abnormal feedback substep, wherein the abnormal feedback substep comprises the step of storing data which cannot be trained and has no training value into the abnormal database when the obtained data are trained one by one.
6. The blockchain application method for the acquisition of the training data set of the robot navigation algorithm according to claim 2, wherein the blockchain application method comprises the following steps: the acquisition interval time is consistent with the time interval generated by the front block and the rear block.
7. The blockchain application method for the acquisition of the training data set of the robot navigation algorithm according to claim 3, wherein the blockchain application method comprises the following steps: in the data uplink substep, data is uploaded to the public application node through an HTTP protocol, and in the data warehousing substep, data is acquired from a corresponding account book of the public application node in an HTTP Restful mode.
8. The blockchain application method for the acquisition of the training data set of the robot navigation algorithm according to claim 2, wherein the blockchain application method comprises the following steps: the preset data type is that the transaction type of the data is set as: and M.
9. The method of claim 1, wherein the method comprises: the data training step includes training the data under a TensorFlow framework.
10. The method of claim 1, wherein the method comprises: the local application database is a non-relational database.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110685237.8A CN113420017B (en) | 2021-06-21 | 2021-06-21 | Block chain application method for acquiring training data set of robot navigation algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110685237.8A CN113420017B (en) | 2021-06-21 | 2021-06-21 | Block chain application method for acquiring training data set of robot navigation algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113420017A true CN113420017A (en) | 2021-09-21 |
| CN113420017B CN113420017B (en) | 2023-10-13 |
Family
ID=77789459
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110685237.8A Active CN113420017B (en) | 2021-06-21 | 2021-06-21 | Block chain application method for acquiring training data set of robot navigation algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113420017B (en) |
Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104252626A (en) * | 2013-06-28 | 2014-12-31 | 康耐视公司 | Semi-supervised method for training multiple pattern recognition and registration tool models |
| CN109542979A (en) * | 2018-11-19 | 2019-03-29 | 广州矩阵信息科技有限公司 | A kind of block catenary system Fast synchronization and the mode of simple smart data storage |
| CN110020774A (en) * | 2019-01-30 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of resource share method, system, device and electronic equipment |
| CN110490305A (en) * | 2019-08-22 | 2019-11-22 | 腾讯科技（深圳）有限公司 | Machine learning model processing method and node based on block chain network |
| CN110610093A (en) * | 2018-06-15 | 2019-12-24 | 国际商业机器公司 | Distributed training in parametric datasets |
| CN110619317A (en) * | 2019-09-26 | 2019-12-27 | 联想(北京)有限公司 | Model training method, model training device and electronic equipment |
| CN110879827A (en) * | 2019-10-23 | 2020-03-13 | 联想(北京)有限公司 | Information processing method and device based on block chain network |
| CN110990871A (en) * | 2019-11-29 | 2020-04-10 | 腾讯云计算（北京）有限责任公司 | Machine learning model training method, prediction method and device based on artificial intelligence |
| CN111368319A (en) * | 2020-03-04 | 2020-07-03 | 西安电子科技大学 | Block chain-based data security access method in federated learning environment |
| US20200272945A1 (en) * | 2019-02-21 | 2020-08-27 | Hewlett Packard Enterprise Development Lp | System and method of decentralized model building for machine learning and data privacy preserving using blockchain |
| CN112269791A (en) * | 2020-11-30 | 2021-01-26 | 上海特高信息技术有限公司 | Construction method of account book index skip list and block chain account book processing method using same |
| CN112445845A (en) * | 2020-12-02 | 2021-03-05 | 平安科技（深圳）有限公司 | Model deployment method, device, equipment and storage medium based on big data mining |
| CN112463889A (en) * | 2020-12-08 | 2021-03-09 | 北京众享比特科技有限公司 | Block chain data storage method, device, equipment and medium |
| CN112529752A (en) * | 2020-12-10 | 2021-03-19 | 清研灵智信息咨询(北京)有限公司 | Trusty investigation platform based on block chain |
| CN112699112A (en) * | 2020-12-31 | 2021-04-23 | 东莞市盟大商业保理有限公司 | Data mining process sharing method based on block chain technology |
| CN112714106A (en) * | 2020-12-17 | 2021-04-27 | 杭州趣链科技有限公司 | Block chain-based federal learning casual vehicle carrying attack defense method |
| CN112783976A (en) * | 2021-01-05 | 2021-05-11 | 上海特高信息技术有限公司 | Consensus algorithm for flexibly packaging out blocks by allied block chains |
| CN112785260A (en) * | 2021-01-20 | 2021-05-11 | 上海特高信息技术有限公司 | Industrial intelligence networking block chain platform |
-
2021
- 2021-06-21 CN CN202110685237.8A patent/CN113420017B/en active Active
Patent Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104252626A (en) * | 2013-06-28 | 2014-12-31 | 康耐视公司 | Semi-supervised method for training multiple pattern recognition and registration tool models |
| CN110610093A (en) * | 2018-06-15 | 2019-12-24 | 国际商业机器公司 | Distributed training in parametric datasets |
| CN109542979A (en) * | 2018-11-19 | 2019-03-29 | 广州矩阵信息科技有限公司 | A kind of block catenary system Fast synchronization and the mode of simple smart data storage |
| CN110020774A (en) * | 2019-01-30 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of resource share method, system, device and electronic equipment |
| US20200272945A1 (en) * | 2019-02-21 | 2020-08-27 | Hewlett Packard Enterprise Development Lp | System and method of decentralized model building for machine learning and data privacy preserving using blockchain |
| CN110490305A (en) * | 2019-08-22 | 2019-11-22 | 腾讯科技（深圳）有限公司 | Machine learning model processing method and node based on block chain network |
| CN110619317A (en) * | 2019-09-26 | 2019-12-27 | 联想(北京)有限公司 | Model training method, model training device and electronic equipment |
| CN110879827A (en) * | 2019-10-23 | 2020-03-13 | 联想(北京)有限公司 | Information processing method and device based on block chain network |
| CN110990871A (en) * | 2019-11-29 | 2020-04-10 | 腾讯云计算（北京）有限责任公司 | Machine learning model training method, prediction method and device based on artificial intelligence |
| CN111368319A (en) * | 2020-03-04 | 2020-07-03 | 西安电子科技大学 | Block chain-based data security access method in federated learning environment |
| CN112269791A (en) * | 2020-11-30 | 2021-01-26 | 上海特高信息技术有限公司 | Construction method of account book index skip list and block chain account book processing method using same |
| CN112445845A (en) * | 2020-12-02 | 2021-03-05 | 平安科技（深圳）有限公司 | Model deployment method, device, equipment and storage medium based on big data mining |
| CN112463889A (en) * | 2020-12-08 | 2021-03-09 | 北京众享比特科技有限公司 | Block chain data storage method, device, equipment and medium |
| CN112529752A (en) * | 2020-12-10 | 2021-03-19 | 清研灵智信息咨询(北京)有限公司 | Trusty investigation platform based on block chain |
| CN112714106A (en) * | 2020-12-17 | 2021-04-27 | 杭州趣链科技有限公司 | Block chain-based federal learning casual vehicle carrying attack defense method |
| CN112699112A (en) * | 2020-12-31 | 2021-04-23 | 东莞市盟大商业保理有限公司 | Data mining process sharing method based on block chain technology |
| CN112783976A (en) * | 2021-01-05 | 2021-05-11 | 上海特高信息技术有限公司 | Consensus algorithm for flexibly packaging out blocks by allied block chains |
| CN112785260A (en) * | 2021-01-20 | 2021-05-11 | 上海特高信息技术有限公司 | Industrial intelligence networking block chain platform |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113420017B (en) | 2023-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110555416B (en) | Plant identification method and device | |
| WO2022017163A1 (en) | Image processing method and apparatus, and device and storage medium | |
| CN111258995B (en) | Data processing method, device, storage medium and equipment | |
| CN111489401B (en) | Image color constancy processing method, system, device and storage medium | |
| CN113518256B (en) | Video processing method, video processing device, electronic equipment and computer readable storage medium | |
| US9665773B2 (en) | Searching for events by attendants | |
| CN107992937B (en) | Unstructured data judgment method and device based on deep learning | |
| CN113239875B (en) | Method, system and device for acquiring face characteristics and computer readable storage medium | |
| CN103838834A (en) | Mentioning recommending method, information processing method and system | |
| CN111898735A (en) | Distillation learning method, distillation learning device, computer equipment and storage medium | |
| WO2018036286A1 (en) | Target-object identification method and apparatus, and robot | |
| CN112001274A (en) | Crowd density determination method, device, storage medium and processor | |
| CN114005012A (en) | Training method, device, equipment and storage medium of multi-mode pre-training model | |
| CN107145879A (en) | A kind of floristics automatic identifying method and system | |
| CN111414842B (en) | Video comparison method and device, computer equipment and storage medium | |
| CN105631404B (en) | The method and device that photo is clustered | |
| CN107291863A (en) | A kind of quick check method and device for industrial control equipment information security | |
| CN112748941A (en) | Feedback information-based target application program updating method and device | |
| CN113573044B (en) | Video data processing method and device, computer equipment and readable storage medium | |
| CN111310516A (en) | Behavior identification method and device | |
| CN105144193A (en) | A method and apparatus for estimating a pose of an imaging device | |
| CN113420017A (en) | Block chain application method for acquiring robot navigation algorithm training data set | |
| CN106714099A (en) | Photograph information processing and scenic spot identification method, client and server | |
| CN112541891B (en) | Crowd counting method and system based on hole convolution high-resolution network | |
| CN112990076A (en) | Data arrangement method and device based on artificial intelligence |
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 |