CN113099175B - Multi-model handheld cloud detection transmission system and detection method based on 5G - Google Patents
Multi-model handheld cloud detection transmission system and detection method based on 5G Download PDFInfo
- Publication number
- CN113099175B CN113099175B CN202110334258.5A CN202110334258A CN113099175B CN 113099175 B CN113099175 B CN 113099175B CN 202110334258 A CN202110334258 A CN 202110334258A CN 113099175 B CN113099175 B CN 113099175B
- Authority
- CN
- China
- Prior art keywords
- detection
- module
- model
- network
- data
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/94—Hardware or software architectures specially adapted for image or video understanding
- G06V10/95—Hardware or software architectures specially adapted for image or video understanding structured as a network, e.g. client-server architectures
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Informatics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a multi-model handheld cloud detection transmission system and a detection method based on 5G, wherein the multi-model handheld cloud detection transmission system based on 5G comprises a handheld device and a cloud detection module, wherein the handheld device is connected with a 5G network and is connected with the cloud detection module through the 5G network; the handheld device is internally provided with an acquisition module, a processing module, a storage module, a detection module and a communication module, wherein the acquisition module is connected with the processing module, the processing module is connected with the detection module, the storage module and the communication module, the detection module is connected with the storage module and the communication module, and the communication module is connected with a 5G network. According to the invention, the handheld device receives the student network detection model for detection in real time through the 5G network, the cloud detection module trains and updates the teacher network detection model through the knowledge distillation module, and the teacher network detection model finishes the training and updating of the student network detection model, so that multiple devices can work simultaneously, and the transmission and processing efficiency is improved.
Description
Technical Field
The invention relates to the technical field of high-speed rail detection machine vision, in particular to a multi-model handheld cloud detection transmission system and a detection method based on 5G.
Background
In recent years, with the continuous development of high-speed rail technology, great convenience is brought to people going out, however, high-speed rails need great manpower and material resources to maintain the safe operation of the high-speed rails, the state of the high-speed rails needs to be detected in advance or in real time in multiple aspects so as to facilitate the safe operation of the high-speed rails, and for some special scenes, people are difficult to observe in real time, and the observation needs to be assisted in an intelligent manner. The detection is particularly important when the high-speed train stops and finishes, and the whole train can be sufficiently ensured so as to be safely operated at the next time. The machine vision system is used as the 'eye' of the robot, and can acquire the image of the detected object by means of an industrial camera, extract effective information from the image through a computer and analyze the effective information, and further realize automatic detection and evaluation of the train.
However, the conventional method usually adopts two schemes of hardware processing and image acquisition in real time. For a real-time fault detection system, because the computing capability of a handheld device is limited, a handheld device can only detect a limited number of faults in general, but cannot detect all-around faults. Meanwhile, the equipment is difficult to upgrade, needs to be returned to developers for system upgrade, has relatively high upgrade and maintenance cost, and is not suitable for the requirement of multiple scenes. For those devices that can update the model through the detection result, a high-end chip is usually required, and the price is also high. However, the method needs a certain manual transmission process, and needs a long time to give a corresponding detection result, so that the efficiency is low, and sometimes the method is limited by the network, and some special places cannot be deployed.
Disclosure of Invention
In order to solve the problems, the invention provides a multi-model handheld cloud detection transmission system and a detection method based on 5G, wherein a handheld device receives a student network detection model in real time through a 5G network for detection, a cloud detection module trains and updates a teacher network detection model through a knowledge distillation module, the teacher network detection model trains and updates the student network detection model, multiple devices can work simultaneously, and transmission and processing efficiency is improved.
The technical scheme adopted by the invention is as follows:
the application provides a multi-model handheld cloud detection transmission system based on 5G, which comprises handheld equipment and a cloud detection module, wherein the handheld equipment is connected with a 5G network and is connected with the cloud detection module through the 5G network;
the handheld device is internally provided with an acquisition module, a processing module, a storage module, a detection module and a communication module, wherein the acquisition module is connected with the processing module, the processing module is connected with the detection module, the storage module and the communication module, the detection module is connected with the storage module and the communication module, and the communication module is connected with a 5G network;
the cloud detection module comprises a database, a knowledge distillation module, a detection unit, a transceiver module and a data processing unit, the knowledge distillation module is connected with the transceiver module, the database and the detection unit, the data processing unit is connected with the transceiver module and the database, the database is further connected with the detection unit and the transceiver module, and the detection unit is connected with the transceiver module.
Preferably, the knowledge distillation module comprises a sample training unit, a plurality of teacher network detection models and a plurality of student network detection models, and the teacher network detection models are used for completing the training and updating of new teacher network detection models and the training of student network detection models through the sample training unit.
Preferably, the model size of the student network detection model is equal to or less than 50M.
Preferably, the acquisition module includes a camera for acquiring data of the scene information.
Based on the above multi-model handheld cloud detection and transmission system based on 5G, the application also provides a detection method using the above multi-model handheld cloud detection and transmission system based on 5G, which includes the following steps:
s1, transmitting scene information to be detected by the handheld device to a transceiver module through a communication module and a 5G network, transmitting the received scene information to a data processing module by the transceiver module for processing, downloading a corresponding trained student network detection model from a database after the data processing module acquires a corresponding scene, and transmitting the corresponding trained student network detection model to the handheld device through the transceiver module;
s2, the handheld device completes initialization after receiving the student network detection model, performs data acquisition through a camera in the acquisition module, and completes fault detection through the student network detection model in the handheld device;
s3, after the handheld device completes detection, integrating detection results of a plurality of parts of the train to obtain corresponding overall detection data, uploading the fault image and the overall detection data to a cloud detection module through a 5G network, and storing the fault image and the overall detection data in a database;
and S4, the cloud detection module performs comprehensive analysis according to the overall detection data to complete training and updating of a new teacher network detection model, and a student network detection model is trained through the knowledge distillation module based on the original data and the teacher network detection model and is used for subsequent detection.
Preferentially, the step S4 specifically includes the following steps:
(41) Based on the detection data in the database, the original teacher network detection model and the original student network detection model, artificial calibration is carried out to be used as a training sample, and the self-supervision-based knowledge distillation loss L are calculated SS The calculation formula of (c) is as follows:
wherein the content of the first and second substances,for the output result after the teacher network detection model is subjected to self-supervision and classification,the method is characterized in that tau is a distillation coefficient, tau =2, x is a positive integer, and j is a positive integer, and is an output result of a student network detection model after self-supervision classification;
(42) Calculating the distillation loss of the common knowledge based on the detection data in the database, the original teacher network detection model and the original student network detection model kd The calculation formula of (a) is as follows:
wherein the content of the first and second substances,for the output result of the teacher network detection model after softmax classification,the output result of the student network detection model after being classified by softmax is shown, C is the number of classes, and x is obeyed by D x ;
(43) And carrying out weighted average on the common knowledge distillation loss and the knowledge distillation loss based on self-supervision, and judging the angle of data by adopting a rotation angle-based mode to serve as a classification task to finish the training and updating of the student network detection model.
The invention has the beneficial effects that:
1. a knowledge distillation module is arranged in the cloud detection module, wherein the teacher network detection model performs weighted average through self-supervision-based knowledge distillation and common knowledge distillation to finish training and updating of the student network detection model, and the detection precision of the student network detection model is improved through refining parameters;
2. the student network detection model that the update was accomplished issues to handheld device through the 5G network and detects, and the model is less, slows down handheld device's operational capability, and a plurality of equipment simultaneous workings of being convenient for slow down transmission pressure and improve detection efficiency.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is an overall block diagram of the system of the present invention;
FIG. 2 is a schematic diagram of the connection of the handheld device of the present invention;
fig. 3 is a schematic connection diagram of the cloud detection module according to the present invention.
The reference numbers in the figures are: 1. the system comprises a handheld device, 11 acquisition modules, 12 processing modules, 13 storage modules, 14 detection modules, 15 communication modules, 2 cloud detection modules, 21 databases, 22 knowledge distillation modules, 221 teacher network detection models, 222 student network detection models, 23 detection units, 24 transceiving modules, 25 data processing units and 3.5G networks.
Detailed Description
As shown in fig. 1, the application provides a multi-model handheld cloud detection transmission system based on 5G, which includes a handheld device 1 and a cloud detection module 2, wherein the handheld device 1 is connected with a 5G network 3 and connected with the cloud detection module 2 through the 5G network 3.
As shown in fig. 2, a collection module 11, a processing module 12, a storage module 13, a detection module 14 and a communication module 15 are arranged in the handheld device 1, the collection module 11 is connected with the processing module 12, the processing module 12 is connected with the detection module 14, the storage module 13 and the communication module 15, the detection module 14 is connected with the storage module 13 and the communication module 15, and the communication module 15 is connected with the 5G network 3. The acquisition module 11 includes a camera for acquiring data of scene information.
As shown in fig. 3, the cloud detection module 2 includes a database 21, a knowledge distillation module 22, a detection unit 23, a transceiver module 24 and a data processing unit 25, the knowledge distillation module 22 is connected to the transceiver module 24, the database 21 and the detection unit 23, the data processing unit 25 is connected to the transceiver module 24 and the database 21, the database 21 is further connected to the detection unit 23 and the transceiver module 24, and the detection unit 23 is connected to the transceiver module 24. The knowledge distilling module 22 comprises a sample training unit, a plurality of teacher network detection models 221 and a plurality of student network detection models 222, wherein the teacher network detection models 221 are used for completing the training and updating of the new teacher network detection models 221 and the training of the student network detection models 222 through the sample training unit. The model size of the student network detection model 222 is 50M or less, the transmission rate of the student network detection model 222 is generally in the order of seconds due to the strong broadband of the 5G network 3, and the amount of parameters used in detection is smaller than that of the teacher network detection model 221.
As shown in fig. 1, based on the above-mentioned multi-model handheld cloud detection transmission system based on 5G, the present application also provides a detection method using the above-mentioned multi-model handheld cloud detection transmission system based on 5G, which includes the following steps:
s1, transmitting scene information detected by the handheld device 1 according to needs to a transceiving module 24 through a communication module 15 and a 5G network 3, transmitting the received scene information to a data processing module 12 through the transceiving module 24 for processing, downloading a corresponding trained student network detection model 222 from a database 21 after the data processing module 12 obtains a corresponding scene, and transmitting the corresponding trained student network detection model to the handheld device 1 through the transceiving module 24.
And S2, the handheld device 1 completes initialization after receiving the student network detection model 222, performs data acquisition through a camera in the acquisition module 11, and completes fault detection through the student network detection model 222 in the handheld device 1.
And S3, after the handheld device 1 completes detection, the detection results of a plurality of parts of the train are integrated to obtain corresponding overall detection data, and the fault image and the overall detection data are uploaded to the cloud detection module 2 through the 5G network 3 and stored in the database 21.
And S4, the cloud detection module 2 performs comprehensive analysis according to the overall detection data to complete the training and updating of the new teacher network detection model 221, and trains out the student network detection model 222 for subsequent detection through the knowledge distillation module 22 based on the original data and the teacher network detection model 221.
The step S4 specifically includes the following steps:
(41) Based on the detection data in the database 21, the original teacher network detection model 221 and the student network detection model 222, artificial calibration is performed as a training sample, and the knowledge distillation loss based on self-supervision and the knowledge distillation loss L based on self-supervision are calculated SS The calculation formula of (a) is as follows:
wherein, the first and the second end of the pipe are connected with each other,for the output result of the teacher network detection model 221 after the self-supervision classification,for the output result of the student network detection model 222 after the self-supervision classification, τ is a distillation coefficient, generally τ =2, x is a positive integer, and j is a positive integer.
(42) Calculating the distillation loss of the common knowledge and the distillation loss L of the common knowledge based on the detection data in the database 21, the original teacher network detection model 221 and the student network detection model 222 kd The calculation formula of (c) is as follows:
wherein the content of the first and second substances,for the output result of the teacher network detection model 221 after softmax classification,the output result of the student network detection model 222 after being sorted by softmax, C is the number of categories, x obeys D x 。
(43) The common knowledge distillation loss and the knowledge distillation loss based on self-supervision are weighted and averaged, and the angle of data is judged in a rotation angle-based mode to serve as a classification task, so that the training and updating of the student network detection model 222 are completed.
Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described above, or equivalents may be substituted for elements thereof. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. The utility model provides a handheld high in clouds detection transmission system of multimode based on 5G which characterized in that: the system comprises a handheld device and a cloud detection module, wherein the handheld device is connected with a 5G network and is connected with the cloud detection module through the 5G network;
the handheld device is internally provided with an acquisition module, a processing module, a storage module, a detection module and a communication module, wherein the acquisition module is connected with the processing module, the processing module is connected with the detection module, the storage module and the communication module, the detection module is connected with the storage module and the communication module, and the communication module is connected with a 5G network;
the cloud detection module comprises a database, a knowledge distillation module, a detection unit, a transceiving module and a data processing unit, the knowledge distillation module is connected with the transceiving module, the database and the detection unit, the data processing unit is connected with the transceiving module and the database, the database is also connected with the detection unit and the transceiving module, and the detection unit is connected with the transceiving module;
the handheld device transmits scene information detected according to needs to a transceiving module through a communication module and a 5G network, the transceiving module transmits the received scene information to a data processing module for processing, and the data processing module downloads a corresponding trained student network detection model from a database after acquiring a corresponding scene and transmits the corresponding trained student network detection model to the handheld device through the transceiving module;
the handheld device completes initialization after receiving the student network detection model, performs data acquisition through a camera in the acquisition module, and completes fault detection through the student network detection model in the handheld device;
after the handheld device completes detection, the detection results of a plurality of parts of the train are integrated to obtain corresponding overall detection data, and the fault image and the overall detection data are uploaded to a cloud detection module through a 5G network and stored in a database;
the cloud detection module performs comprehensive analysis according to the overall detection data to complete the training and updating of a new teacher network detection model, and the student network detection model is trained through the knowledge distillation module based on the original data and the teacher network detection model for subsequent detection.
2. The 5G-based multi-model handheld cloud detection transmission system according to claim 1, wherein: the knowledge distillation module comprises a sample training unit, a plurality of teacher network detection models and a plurality of student network detection models, wherein the teacher network detection models are used for completing the training and updating of new teacher network detection models and the training of the student network detection models through the sample training unit.
3. The 5G-based multi-model handheld cloud detection transmission system according to claim 2, wherein: the size of the model of the student network detection model is less than or equal to 50M.
4. The 5G-based multi-model handheld cloud detection transmission system according to claim 3, wherein: the acquisition module comprises a camera and is used for acquiring scene information data.
5. A detection method using the 5G-based multi-model handheld cloud detection transmission system according to claim 4, wherein: the method comprises the following steps:
s1, transmitting scene information to be detected by the handheld device to a transceiver module through a communication module and a 5G network, transmitting the received scene information to a data processing module by the transceiver module for processing, downloading a corresponding trained student network detection model from a database after the data processing module acquires a corresponding scene, and transmitting the corresponding trained student network detection model to the handheld device through the transceiver module;
s2, the handheld device completes initialization after receiving the student network detection model, performs data acquisition through a camera in the acquisition module, and completes fault detection through the student network detection model in the handheld device;
s3, after the handheld device completes detection, integrating detection results of a plurality of parts of the train to obtain corresponding overall detection data, uploading the fault image and the overall detection data to a cloud detection module through a 5G network, and storing the fault image and the overall detection data in a database;
and S4, the cloud detection module performs comprehensive analysis according to the overall detection data to complete training and updating of a new teacher network detection model, and a student network detection model is trained through the knowledge distillation module based on the original data and the teacher network detection model and is used for subsequent detection.
6. The detection method of the 5G-based multi-model handheld cloud detection transmission system according to claim 5, wherein: step S4 specifically includes the following steps:
(41) Based on the detection data in the database, the original teacher network detection model and the original student network detection model, artificial calibration is carried out to be used as a training sample, and the knowledge distillation loss based on self-supervision and the knowledge distillation loss L based on self-supervision are calculated SS The calculation formula of (a) is as follows:
wherein the content of the first and second substances,the output result of the teacher network detection model after self-supervision and classification is obtained,the method is characterized in that tau is a distillation coefficient, tau =2, x is a positive integer, and j is a positive integer, and is an output result of a student network detection model after self-supervision classification; i is the serial number of the identification category, s is the student, and t is the teacher;
(42) Calculating the distillation loss of the common knowledge based on the detection data in the database, the original teacher network detection model and the original student network detection model kd Is calculated as followsShown in the figure:
wherein, the first and the second end of the pipe are connected with each other,the output result of the teacher network detection model after being classified by softmax,the output result of the student network detection model after softmax classification is obtained, C is the number of classes, and x obeys Dx;
(43) And carrying out weighted average on the common knowledge distillation loss and the knowledge distillation loss based on self-supervision, and judging the angle of data by adopting a rotation angle-based mode to serve as a classification task to finish training and updating of a student network detection model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110334258.5A CN113099175B (en) | 2021-03-29 | 2021-03-29 | Multi-model handheld cloud detection transmission system and detection method based on 5G |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110334258.5A CN113099175B (en) | 2021-03-29 | 2021-03-29 | Multi-model handheld cloud detection transmission system and detection method based on 5G |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113099175A CN113099175A (en) | 2021-07-09 |
CN113099175B true CN113099175B (en) | 2022-11-04 |
Family
ID=76670736
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110334258.5A Active CN113099175B (en) | 2021-03-29 | 2021-03-29 | Multi-model handheld cloud detection transmission system and detection method based on 5G |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113099175B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118154992A (en) * | 2024-05-09 | 2024-06-07 | 中国科学技术大学 | Medical image classification method, device and storage medium based on knowledge distillation |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111899846B (en) * | 2020-07-14 | 2023-07-25 | 苏州睿达科技有限公司 | Tongue image detector based on end cloud cooperation and detection method thereof |
CN112070207A (en) * | 2020-07-31 | 2020-12-11 | 华为技术有限公司 | Model training method and device |
CN111709409B (en) * | 2020-08-20 | 2020-11-20 | 腾讯科技(深圳)有限公司 | Face living body detection method, device, equipment and medium |
CN112241455B (en) * | 2020-12-17 | 2021-05-04 | 之江实验室 | Automatic compression method and platform based on multi-level knowledge distillation pre-training language model |
-
2021
- 2021-03-29 CN CN202110334258.5A patent/CN113099175B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN113099175A (en) | 2021-07-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109389161B (en) | Garbage identification evolutionary learning method, device, system and medium based on deep learning | |
CN103324937B (en) | The method and apparatus of label target | |
CN111974704A (en) | Garbage classification detection system and method based on computer vision | |
CN107369635B (en) | Intelligent semiconductor equipment system based on deep learning | |
CN102053563A (en) | Flight training data acquisition and quality evaluation system of analog machine | |
CN106251695A (en) | Destination's parking stall intelligent recommendation system and method based on parking space state monitoring | |
CN110046584B (en) | Road crack detection device and detection method based on unmanned aerial vehicle inspection | |
CN113780435B (en) | Vehicle damage detection method, device, equipment and storage medium | |
CN108672316A (en) | A kind of micro parts quality detecting system based on convolutional neural networks | |
CN113099175B (en) | Multi-model handheld cloud detection transmission system and detection method based on 5G | |
CN105184229A (en) | Online learning based real-time pedestrian detection method in dynamic scene | |
CN111160432A (en) | Automatic classification method and system for panel production defects | |
CN112613438A (en) | Portable online citrus yield measuring instrument | |
CN110910440A (en) | Power transmission line length determination method and system based on power image data | |
CN115100554A (en) | Unmanned aerial vehicle power inspection system based on intelligent vision and detection method thereof | |
CN111735815A (en) | Method and device for detecting defects of small hardware fittings of power transmission line and storage medium | |
CN112788110A (en) | Product appearance detection method based on cloud edge collaborative model optimization and implementation system thereof | |
CN110490166A (en) | High iron catenary pipe cap automatic testing method based on depth learning technology | |
CN114494916A (en) | Black-neck crane monitoring and tracking method based on YOLO and DeepsORT | |
CN116128879B (en) | Lightweight transmission line defect detection method and device | |
CN112819988A (en) | Unmanned aerial vehicle power station intelligent inspection method and system based on 5G and network side server | |
CN116206148A (en) | Intelligent detection, identification and classification method for decoration waste | |
CN112418362B (en) | Target detection training sample screening method | |
CN114445411A (en) | Unmanned aerial vehicle line patrol defect identification system and control method | |
WO2022241597A1 (en) | Ai intelligent garbage identification and classification 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |