CN113612966A - Intelligent specific target object distinguishing method based on urban low-altitude remote sensing data - Google Patents
Intelligent specific target object distinguishing method based on urban low-altitude remote sensing data Download PDFInfo
- Publication number
- CN113612966A CN113612966A CN202110753189.1A CN202110753189A CN113612966A CN 113612966 A CN113612966 A CN 113612966A CN 202110753189 A CN202110753189 A CN 202110753189A CN 113612966 A CN113612966 A CN 113612966A
- Authority
- CN
- China
- Prior art keywords
- intelligent
- aerial vehicle
- unmanned aerial
- server
- remote sensing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000012545 processing Methods 0.000 claims abstract description 34
- 238000004891 communication Methods 0.000 claims abstract description 27
- 230000011664 signaling Effects 0.000 claims description 14
- 238000013473 artificial intelligence Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 8
- 230000006835 compression Effects 0.000 claims description 6
- 238000007906 compression Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000012850 discrimination method Methods 0.000 claims 1
- 230000005540 biological transmission Effects 0.000 description 11
- 238000004364 calculation method Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008309 brain mechanism Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
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
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
- H04L65/60—Network streaming of media packets
- H04L65/65—Network streaming protocols, e.g. real-time transport protocol [RTP] or real-time control protocol [RTCP]
-
- 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
- H04L67/10—Protocols in which an application is distributed across nodes in the network
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Networks & Wireless Communication (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a specific target object intelligent distinguishing method based on urban low-altitude remote sensing data, which comprises a platform system, a server, a camera, a moving end, a holder, a universal interface, an unmanned aerial vehicle, an intelligent processing box, a 5G communication module, an intelligent graphic processing chip, a video plug-flow module and an intelligent identification module, wherein the platform system is communicated with the unmanned aerial vehicle through the server, and the unmanned aerial vehicle is communicated with the moving end through the 5G communication module; the method comprises the following steps: step 1, adopting high-definition image data through an unmanned aerial vehicle, wherein the resolution of the data is unlimited, and before the data is marked, renaming is carried out to enable the name to be more standard; the invention has the advantages of high intelligent degree, convenient control, strong practicability, reduced identification time, reduced delay and convenient popularization and use.
Description
Technical Field
The invention relates to the technical field of remote sensing, in particular to an intelligent specific target object distinguishing method based on urban low-altitude remote sensing data.
Background
The unmanned aircraft is an unmanned aerial vehicle which is operated by utilizing a radio remote control device and a self-contained program control device, the development of the unmanned aerial vehicle technology and the continuous reduction of the cost of key components greatly reduce the threshold for developing an unmanned aerial vehicle system, the unmanned aerial vehicle is widely applied, the current development wave of the artificial intelligence industry mainly comes from the proposal of a deep learning algorithm, large-scale calculation is realized on the basis of data quantity and calculation capacity, and the unmanned aerial vehicle belongs to technical breakthrough. Belongs to super artificial intelligence, and has room for continuous breakthrough in the basic theory research on the aspects of consciousness origin, human brain mechanism and the like;
at present, the acquisition of urban low-altitude remote sensing data is mostly completed by adopting an unmanned aerial vehicle, the image transmission of the unmanned aerial vehicle adopts radio transmission, the radio transmission is greatly influenced by the distance between the unmanned aerial vehicle and a ground station, the bandwidth of the radio is limited, so that the unmanned aerial vehicle reduces the image quality of an original video in order to meet the fluency of the video when a video image is transmitted, the video resolution is always kept at 1080P, but the requirement on the video definition is higher and higher in practical application, 1080P cannot meet the current requirement, in the aspect of artificial intelligent identification, an artificial intelligent identification method aiming at the unmanned aerial vehicle while in-flight identification is relatively immature, the identification time is relatively long, and obvious delay is generated when the artificial intelligent identification method is superposed on the video, so how to adopt a 5G network to replace the radio to transmit image data and flight data, how to train a light-weight identification model through an optimization algorithm, the identification time is shortened, the image quality of unmanned aerial vehicle image transmission is improved while the smoothness of a video is guaranteed, the problem that the 1080P bottleneck is required to be solved at present is broken through, and therefore an intelligent specific target object distinguishing method based on urban low-altitude remote sensing data is provided.
Disclosure of Invention
The invention aims to provide the intelligent specific target object distinguishing method based on the urban low-altitude remote sensing data, which has the advantages of high intelligent degree, convenience in control, strong practicability, recognition time reduction, delay reduction and convenience in popularization and use, and solves the problems in the prior art.
The above object of the present invention is achieved by the following technical solutions: a specific target object intelligent distinguishing method based on urban low-altitude remote sensing data comprises a platform system, a server, a camera, a moving end, a holder, a universal interface, an unmanned aerial vehicle, an intelligent processing box, a 5G communication module, an intelligent graphic processing chip, a video plug-flow module and an intelligent identification module, wherein the platform system is communicated with the unmanned aerial vehicle through the server, and the unmanned aerial vehicle is communicated with the moving end through the 5G communication module;
the method comprises the following steps:
step 1, high-definition image data is adopted through the unmanned aerial vehicle, the resolution of the data is unlimited, and names are renamed to be more standard before the data are marked.
And 2, configuring the environment of a marking tool, marking the preprocessed image data by using the marking tool, converting the output file into a file required by the model, and sorting the data into a mode approved by the model.
And 3, downloading a pre-training weight file, training a lightweight model by adopting a framework integrating two functions of object detection and instance segmentation, adjusting parameters and optimizing a recognition algorithm.
And 4, testing the trained model, and selecting the optimal model.
And 5, writing an artificial intelligence recognition program by using the model in the step 4.
And 6, deploying signaling service on the server, configuring the camera to the server, remotely controlling the camera holder by calling an interface, and acquiring 4K high-definition image data.
And 7, building a mobile terminal APP, controlling the unmanned aerial vehicle and the cradle head at the mobile terminal, and meanwhile, receiving a 4K high-definition real-time picture to acquire real-time flight data of the unmanned aerial vehicle.
And 8, building a platform system for watching the real-time discrimination result of the high-definition image data of the unmanned aerial vehicle and the returned flight data.
The invention is further configured to: the intelligent processing box is composed of a 5G communication module, an intelligent graphic processing chip, a video plug-flow module and an intelligent identification module.
By adopting the technical scheme, the intelligent processing box directly identifies the video after acquiring the video in an edge calculation mode, and then carries out live streaming on the identified video through the video streaming module, wherein the video streaming supports rtmp and rtsp protocols, so as to meet the application requirement of real-time performance.
The invention is further configured to: in the step 4, the intelligent recognition module acquires pictures through unmanned aerial vehicle aerial photography, trains a recognition model by using an object instance segmentation framework, and obtains an optimal lightweight model through continuous parameter adjustment and optimization.
By adopting the technical scheme, the processing mode of the lightweight identification model is adopted, the service of the target detection is deployed at one end close to the data source, the network service can be responded more quickly, and the identification time is reduced to a greater extent by using the lightweight identification model.
The invention is further configured to: the intelligent image processing chip adopts an efficient lossless video compression algorithm to perform lossless video compression on the original video stream acquired by the camera, and supports H.265 and H.264 coding protocols.
By adopting the technical scheme, the graphic information can be processed and transmitted more quickly, and the video acquisition efficiency is improved.
The invention is further configured to: and the server deploys signaling service, configures the camera information assembled by the unmanned aerial vehicle on the server, and performs remote pan-tilt control by calling an interface on the server.
Through adopting above-mentioned technical scheme, make things convenient for operating personnel long-range to control unmanned aerial vehicle to in order to carry out video acquisition to the environment under the different situation for the people.
The invention is further configured to: the mobile terminal is a mobile phone or a tablet personal computer, and the mobile terminal controls the cloud deck of the camera in real time by calling the interface of the server and receives flight data transmitted back from the unmanned aerial vehicle in real time.
Through adopting above-mentioned technical scheme for ground staff can be real-time the video information of receiving unmanned aerial vehicle collection, the remote control of being convenient for simultaneously.
The invention is further configured to: in the step 3, the recognition algorithm integrates two functions of object detection and example segmentation, and a light-weight deep separable convolutional neural network is adopted to train an ultra-light-weight recognition model.
By adopting the technical scheme, the network service can be responded more quickly, and the lightweight identification model is used for reducing the identification time to a greater extent.
The invention is further configured to: the camera is in communication connection with the platform system through the 5G communication module.
By adopting the technical scheme, the platform system can be set up to watch the high-definition identification image of the unmanned aerial vehicle and the acquired flight data of the unmanned aerial vehicle, and draw the identified area and the identified type on the map in real time.
In conclusion, the beneficial technical effects of the invention are as follows:
1. the intelligent specific target object distinguishing method based on urban low-altitude remote sensing data structurally and effectively breaks through the image transmission bottleneck of 1080P of an unmanned aerial vehicle through the edge calculation scheme of a 5G communication module, a server and an intelligent processing box, improves the image quality to 4K while ensuring the image transmission smoothness of the unmanned aerial vehicle, realizes real-time control of a remote holder by using signaling service, places artificial intelligent identification service at one end close to a data source, provides nearest service nearby, generates faster network service response, uses an ultra-light-weight identification model at the same time, meets the basic requirements of real-time service and application intelligence, and can watch real-time identification images and flight data through a platform system;
2. structurally, unmanned aerial vehicle is through carrying on high definition digtal camera, the 4K image that unmanned aerial vehicle gathered passes back through the 5G network in real time, support many screens, the multichannel is broadcast simultaneously, and realize the long-range cloud platform control function of mobile terminal to the camera through signaling service, traditional unmanned aerial vehicle video picture pass definition and cloud platform control are removed effectively and are received the distance restriction of unmanned aerial vehicle and ground satellite station when improving video definition, attached the AI service based on the high definition image simultaneously, adopt edge calculation and ultra light weight level recognition model, unmanned aerial vehicle 4K high definition video live broadcast and artificial intelligence discernment's high efficiency has been realized and has been fused.
Drawings
FIG. 1 is a schematic structural diagram of a working principle of an intelligent specific target object distinguishing method based on urban low-altitude remote sensing data according to the invention;
FIG. 2 is a schematic structural diagram of the method for intelligently distinguishing the specific target object based on the urban low-altitude remote sensing data.
In the figure, 1, a platform system; 2. a server; 3. a camera; 4. a mobile terminal; 5. a holder; 6. a general purpose interface; 7. an unmanned aerial vehicle; 8. an intelligent processing box; 9. 5G communication module; 10. an intelligent graphics processing chip; 11. a video plug-flow module; 12. and (5) an intelligent identification module.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments, and all other embodiments obtained by a person of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the protection scope of the present invention.
In the description of the present invention, it is to be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", and the like, indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are merely for convenience in describing the present invention and simplifying the description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention.
Referring to fig. 1 and 2, the system comprises a platform system 1, a server 2, a camera 3, a mobile terminal 4, a cradle head 5, a universal interface 6, an unmanned aerial vehicle 7, an intelligent processing box 8, a 5G communication module 9, an intelligent graphic processing chip 10, a video plug-flow module 11 and an intelligent identification module 12, wherein the platform system 1 establishes communication with the unmanned aerial vehicle 7 through the server 2, and the unmanned aerial vehicle 7 communicates with the mobile terminal 4 through the 5G communication module 9;
the method comprises the following steps:
step 1, high-definition image data is acquired through the unmanned aerial vehicle 7, the resolution of the data is not limited, and names are more standard by renaming before the data are marked.
And 2, configuring the environment of a marking tool, marking the preprocessed image data by using the marking tool, converting the output file into a file required by the model, and sorting the data into a mode approved by the model.
And 3, downloading a pre-training weight file, training a lightweight model by adopting a framework integrating two functions of object detection and instance segmentation, adjusting parameters and optimizing a recognition algorithm.
And 4, testing the trained model, and selecting the optimal model.
And 5, writing an artificial intelligence recognition program by using the model in the step 4.
Step 6, signaling service is deployed on the server 2, the camera 3 is configured on the server 2, the camera 3 holder 5 is remotely controlled by calling an interface, and 4K high-definition image data is acquired.
Step 7, build and remove end 4APP, control unmanned aerial vehicle 7 and cloud platform 5 at removal end 4, can receive the real-time picture of 4K high definition simultaneously, acquire the real-time flight data of unmanned aerial vehicle 7.
Step 8, the platform system 1 is set up for watching the real-time discrimination result of the high-definition image data of the unmanned aerial vehicle 7 and the returned flight data;
in this embodiment, in the using process, 5G communication module 9, intelligent graphics processing chip 10, video plug-flow module 11 and intelligent identification module 12 are integrated into intelligent processing box 8, intelligent processing box 8 and camera 3 are connected and carried on unmanned aerial vehicle 7, deploy signaling service on server 2, configure camera 3 onto server 2, take high definition image data through unmanned aerial vehicle 7, data resolution is unlimited, the 4K image that unmanned aerial vehicle 7 gathered passes back through the 5G network in real time, support many screens, multichannel simultaneous playing, and realize moving end 4 to the long-range cloud platform 5 control function of camera 3 through signaling service, the video image of traditional unmanned aerial vehicle 7 is passed the definition and cloud platform 5 control is removed effectively and is received the distance restriction of unmanned aerial vehicle 7 and ground station when improving the video definition.
As shown in fig. 1 and fig. 2, the system comprises a platform system 1, a server 2, a camera 3, a mobile terminal 4, a cradle head 5, a universal interface 6, an unmanned aerial vehicle 7, an intelligent processing box 8, a 5G communication module 9, an intelligent graphic processing chip 10, a video plug-flow module 11 and an intelligent identification module 12, wherein the platform system 1 establishes communication with the unmanned aerial vehicle 7 through the server 2, and the unmanned aerial vehicle 7 communicates with the mobile terminal 4 through the 5G communication module 9;
in the embodiment, the high-efficiency fusion of the unmanned aerial vehicle 7 high-definition video live broadcast and artificial intelligence recognition is realized by adopting an edge calculation and ultra-lightweight recognition model, an interface is called to remotely control the pan/tilt head 5 of the camera 3, 4K high-definition image data is acquired, by the edge calculation scheme of the 5G communication module 9, the server 2 and the intelligent processing box 8, the image transmission bottleneck of the unmanned aerial vehicle 71080P is effectively broken, the image quality is improved to 4K while the smoothness of the image transmission of the unmanned aerial vehicle 7 is ensured, the real-time control of the remote pan-tilt 5 is realized by using signaling service, and the artificial intelligence recognition service is placed at one end close to the data source, the nearest end service is provided nearby, faster network service response is generated, meanwhile, the ultra-light recognition model is used, so that the basic requirements of real-time business and application intelligence are met, and real-time recognition images and flight data can be watched through the platform system 1.
Referring to fig. 1, the intelligent processing box 8 is composed of a 5G communication module 9, an intelligent graphic processing chip 10, a video plug-flow module 11 and an intelligent identification module 12, the intelligent processing box 8 directly identifies the video after acquiring the video in an edge calculation mode, then the identified video is plug-flow live broadcast through the video plug-flow module 11, and the video plug-flow supports rtmp and rtsp protocols so as to meet the application requirement of real-time performance.
Referring to fig. 2, in step 4, the intelligent recognition module 12 acquires images through aerial photography by the unmanned aerial vehicle 7, trains a recognition model by using an object instance segmentation framework, obtains an optimal lightweight model through continuous parameter adjustment and optimization, deploys a service for target detection at one end close to a data source by adopting a processing mode of the lightweight recognition model, can respond to a network service more quickly, and reduces recognition time to a greater extent by using the lightweight recognition model.
Referring to fig. 1, the intelligent image processing chip performs lossless video compression on the original video stream acquired by the camera 3 by using an efficient lossless video compression algorithm, supports h.265 and h.264 encoding protocols, can more quickly process and transmit the graphic information, and improves the video acquisition efficiency.
Referring to fig. 1, deploy signaling service on the server 2, dispose 3 information configurations of camera with unmanned aerial vehicle 7 assembly on server 2, thereby carry out long-range cloud platform 5 control through the interface of calling on the server 2, make things convenient for operating personnel long-range controlling unmanned aerial vehicle 7 to in order to artificially carry out video acquisition to the environment under the different situation.
Referring to fig. 1, remove end 4 and be cell-phone or flat board, remove end 4 and carry out real-time control to the cloud platform 5 of camera 3 through the interface of calling server 2, receive the flight data that comes from the real-time passback of unmanned aerial vehicle 7 for ground staff can be real-time receive the video information that unmanned aerial vehicle 7 gathered, the remote control of being convenient for simultaneously.
Referring to fig. 2, in step 3, the recognition algorithm integrates two functions of object detection and instance segmentation, and a lightweight deep separable convolutional neural network is adopted to train an ultra-lightweight recognition model, so that the network service can be responded to more quickly, and meanwhile, the lightweight recognition model is used to reduce the recognition time to a greater extent.
Referring to fig. 1, camera 3 establishes communication connection with platform system 1 through 5G communication module 9, and platform system 1 is built to watch 7 high definition identification images of unmanned aerial vehicle and the 7 flight data of unmanned aerial vehicle who acquires to draw the region and the identification type that discern in real time on the map.
The implementation principle of the embodiment is as follows:
the invention relates to a specific target object intelligent distinguishing method based on urban low-altitude remote sensing data, in the using process, a 5G communication module 9, an intelligent graphic processing chip 10, a video plug-flow module 11 and an intelligent recognition module 12 are integrated into an intelligent processing box 8, the intelligent processing box 8 and a camera 3 are connected and carried on an unmanned aerial vehicle 7, signaling service is deployed on a server 2, the camera 3 is configured on the server 2, high-definition image data is taken through the unmanned aerial vehicle 7, the resolution of the data is unlimited, 4K images collected by the unmanned aerial vehicle 7 are returned in real time through a 5G network to support multi-screen and multi-path simultaneous playing, and the control function of a mobile terminal 4 on a remote tripod head 5 of the camera 3 is realized through the signaling service, the video definition is improved, and the video image transmission definition of the traditional unmanned aerial vehicle 7 and the distance limitation of the unmanned aerial vehicle 7 and a ground station on tripod head 5 control are effectively eliminated, the edge calculation and ultra-light-weight recognition model are adopted to realize the high-efficiency fusion of the unmanned aerial vehicle 7 high-definition video live broadcast and the artificial intelligence recognition, an interface is called to remotely control the pan/tilt head 5 of the camera 3, 4K high-definition image data is obtained, the image transmission bottleneck of the unmanned aerial vehicle 71080P is effectively broken through the edge calculation scheme of the 5G communication module 9, the server 2 and the intelligent processing box 8, the image quality is improved to 4K while the smoothness of the image transmission of the unmanned aerial vehicle 7 is ensured, the remote pan/tilt head 5 is controlled in real time by using signaling service, the artificial intelligence recognition service is placed at one end close to a data source, the nearest service is provided nearby, the faster network service response is generated, meanwhile, the ultra-light-weight recognition model is used, the basic requirements of real-time service and application intelligence are met, and the real-time recognition image and flight data can be watched through the platform system 1, in order to solve the technical problems existing in the prior art.
While there have been shown and described the fundamental principles and essential features of the invention and advantages thereof, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof; the present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein, and any reference signs in the claims are not intended to be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.
Claims (8)
1. The utility model provides a specific target object intelligence discrimination method based on city low latitude remote sensing data, includes platform system (1), server (2), camera (3), removes end (4), cloud platform (5), general interface (6), unmanned aerial vehicle (7), intelligent processing box (8), 5G communication module (9), intelligent graphics processing chip (10), video plug-flow module (11) and intelligent recognition module (12), its characterized in that: the platform system (1) is communicated with the unmanned aerial vehicle (7) through a server (2), and the unmanned aerial vehicle (7) is communicated with the mobile terminal (4) through a 5G communication module (9);
the method comprises the following steps:
step 1, high-definition image data are acquired through an unmanned aerial vehicle (7), the resolution of the data is not limited, and names are renamed to be more standard before the data are marked;
step 2, configuring the environment of a marking tool, marking the preprocessed image data by using the marking tool, converting the output file into a file required by the model, and sorting the data into a mode approved by the model;
step 3, downloading a pre-training weight file, training a lightweight model by adopting a framework integrating two functions of object detection and instance segmentation, adjusting parameters and optimizing a recognition algorithm;
step 4, testing the trained model, and selecting an optimal model;
step 5, compiling an artificial intelligence identification program by using the model in the step 4;
step 6, deploying signaling service on the server (2), configuring the camera (3) on the server (2), remotely controlling a pan-tilt (5) of the camera (3) through a calling interface, and acquiring 4K high-definition image data;
step 7, building an APP at the mobile terminal (4), controlling the unmanned aerial vehicle (7) and the cradle head (5) at the mobile terminal (4), and simultaneously receiving a 4K high-definition real-time image to acquire real-time flight data of the unmanned aerial vehicle (7);
and 8, building a platform system (1) for watching the real-time discrimination result of the high-definition image data of the unmanned aerial vehicle (7) and the returned flight data.
2. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: the intelligent processing box (8) is composed of a 5G communication module (9), an intelligent graphic processing chip (10), a video plug-flow module (11) and an intelligent identification module (12).
3. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: in the step 4, the intelligent recognition module (12) acquires pictures through aerial photography by the unmanned aerial vehicle (7), trains a recognition model by applying an object instance segmentation framework, and obtains an optimal lightweight model through continuous parameter adjustment and optimization.
4. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: the intelligent image processing chip adopts an efficient lossless video compression algorithm to perform lossless video compression on the original video stream acquired by the camera (3), and supports H.265 and H.264 coding protocols.
5. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: the server (2) is deployed with signaling service, information of a camera (3) assembled by the unmanned aerial vehicle (7) is configured on the server (2), and the remote pan-tilt (5) is controlled by calling an interface on the server (2).
6. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: the mobile terminal (4) is a mobile phone or a tablet, the mobile terminal (4) controls the cloud deck (5) of the camera (3) in real time by calling an interface of the server (2), and receives flight data transmitted back from the unmanned aerial vehicle (7) in real time.
7. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: in the step 3, the recognition algorithm integrates two functions of object detection and example segmentation, and a light-weight deep separable convolutional neural network is adopted to train an ultra-light-weight recognition model.
8. The intelligent distinguishing method for the specific target object based on the urban low-altitude remote sensing data according to claim 1, characterized in that: the camera (3) is in communication connection with the platform system (1) through the 5G communication module (9).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110753189.1A CN113612966A (en) | 2021-07-02 | 2021-07-02 | Intelligent specific target object distinguishing method based on urban low-altitude remote sensing data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110753189.1A CN113612966A (en) | 2021-07-02 | 2021-07-02 | Intelligent specific target object distinguishing method based on urban low-altitude remote sensing data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113612966A true CN113612966A (en) | 2021-11-05 |
Family
ID=78337225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110753189.1A Pending CN113612966A (en) | 2021-07-02 | 2021-07-02 | Intelligent specific target object distinguishing method based on urban low-altitude remote sensing data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113612966A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104811674A (en) * | 2015-05-22 | 2015-07-29 | 郑君雄 | Prison intelligent security device based on multi-rotor unmanned aerial vehicle |
| CN111479086A (en) * | 2019-01-04 | 2020-07-31 | 西南林业大学 | Intelligent real-time command management system for unmanned aerial vehicle |
| WO2020221284A1 (en) * | 2019-04-29 | 2020-11-05 | 黄河勘测规划设计研究院有限公司 | Unmanned aerial vehicle monitoring method and system for basin-wide flood scene |
| CN112422783A (en) * | 2020-10-10 | 2021-02-26 | 广东华南水电高新技术开发有限公司 | Unmanned aerial vehicle intelligent patrol system based on parking apron cluster |
| KR102254491B1 (en) * | 2019-11-22 | 2021-05-21 | (주)가시 | Automatic fly drone embedded with intelligent image analysis function |
-
2021
- 2021-07-02 CN CN202110753189.1A patent/CN113612966A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104811674A (en) * | 2015-05-22 | 2015-07-29 | 郑君雄 | Prison intelligent security device based on multi-rotor unmanned aerial vehicle |
| CN111479086A (en) * | 2019-01-04 | 2020-07-31 | 西南林业大学 | Intelligent real-time command management system for unmanned aerial vehicle |
| WO2020221284A1 (en) * | 2019-04-29 | 2020-11-05 | 黄河勘测规划设计研究院有限公司 | Unmanned aerial vehicle monitoring method and system for basin-wide flood scene |
| KR102254491B1 (en) * | 2019-11-22 | 2021-05-21 | (주)가시 | Automatic fly drone embedded with intelligent image analysis function |
| CN112422783A (en) * | 2020-10-10 | 2021-02-26 | 广东华南水电高新技术开发有限公司 | Unmanned aerial vehicle intelligent patrol system based on parking apron cluster |
Non-Patent Citations (1)
| Title |
|---|
| 朱瑞;施麟;陈宁;: "基于移动平台的海域无人机监视监测系统研究", 江苏科技信息, no. 10 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5709906B2 (en) | Augmented reality panorama for the visually impaired | |
| US20230360254A1 (en) | Pose estimation method and related apparatus | |
| CN112581590B (en) | Unmanned aerial vehicle cloud edge terminal cooperative control method for 5G security rescue networking | |
| WO2022141477A1 (en) | Image processing method and device | |
| CN105049812B (en) | A kind of unmanned plane portable type ground station processing method and system | |
| WO2022141418A1 (en) | Image processing method and device | |
| WO2020083114A1 (en) | Image processing method and system for aerial camera, and unmanned aerial vehicle | |
| WO2022141445A1 (en) | Image processing method and device | |
| CN109886096A (en) | A kind of supervision of smart travel and safety emergency management interlinked command system | |
| US10602064B2 (en) | Photographing method and photographing device of unmanned aerial vehicle, unmanned aerial vehicle, and ground control device | |
| WO2022141351A1 (en) | Vision sensor chip, method for operating vision sensor chip, and device | |
| CN104811615A (en) | Motion control camera shooting system and method | |
| CN208366352U (en) | A kind of guide equipment | |
| CN108650494A (en) | The live broadcast system that can obtain high definition photo immediately based on voice control | |
| CN108650522A (en) | Based on the live broadcast system that can obtain high definition photo immediately automatically controlled | |
| CN116134484A (en) | Image processing method and device | |
| CN109522996A (en) | Constant mark object automatic identification control system | |
| CN115170990A (en) | Artificial intelligent edge computing system and method for unmanned aerial vehicle airborne pod | |
| WO2021164082A1 (en) | Video data acquisition controlling method, transmission control apparatus and wireless transmission system | |
| CN115064023A (en) | Portable terminal teaching training system based on AR glasses | |
| CN111294558A (en) | Forest fire prevention method, device and system based on linkage of fire detection radar, camera and unmanned aerial vehicle | |
| CN113612966A (en) | Intelligent specific target object distinguishing method based on urban low-altitude remote sensing data | |
| CN117041496A (en) | Electric power production operation video monitoring system suitable for capital construction site | |
| CN112004060B (en) | Unmanned aerial vehicle sightseeing method for scenic spot | |
| CN113542562A (en) | Double-light fusion unmanned aerial vehicle's image acquisition system |
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 |