CN111460968B - Unmanned aerial vehicle identification and tracking method and device based on video - Google Patents
Unmanned aerial vehicle identification and tracking method and device based on video Download PDFInfo
- Publication number
- CN111460968B CN111460968B CN202010231230.4A CN202010231230A CN111460968B CN 111460968 B CN111460968 B CN 111460968B CN 202010231230 A CN202010231230 A CN 202010231230A CN 111460968 B CN111460968 B CN 111460968B
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- aerial vehicle
- tracking
- video
- frame
- 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
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000001514 detection method Methods 0.000 claims abstract description 43
- 238000013135 deep learning Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 8
- 238000002372 labelling Methods 0.000 claims abstract description 5
- 230000015654 memory Effects 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 18
- 238000004590 computer program Methods 0.000 claims description 8
- 230000007246 mechanism Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 4
- 230000000694 effects Effects 0.000 description 8
- 238000011176 pooling Methods 0.000 description 8
- 230000006872 improvement Effects 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- 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
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a video-based unmanned aerial vehicle identification and tracking method and device, wherein the method comprises the following steps: manually labeling unmanned aerial vehicles on the collected data sets one by one to obtain unmanned aerial vehicle labeling samples with multiple models and different sizes; training a network based on YOLOv3 by using the data set to obtain a trained deep learning target detection model; the image quality of the unmanned aerial vehicle video to be detected is improved by adopting a Retinex image enhancement means, and each frame of the unmanned aerial vehicle video to be detected is identified through a deep learning target detection model; and realizing fast unmanned aerial vehicle tracking in the video based on a Sort algorithm. The method can identify the unmanned aerial vehicle in the video and track the unmanned aerial vehicle with high robustness and high precision, can enhance the image when the image of the unmanned aerial vehicle is unclear, and is suitable for various complex scenes.
Description
Technical Field
The invention relates to the field of unmanned aerial vehicle identification and tracking, in particular to a video-based unmanned aerial vehicle identification and tracking method and device.
Background
The video-based moving target detection and tracking problem has a certain research foundation on scientific and technical development and engineering application, and has more mature solutions in the fields of intelligent transportation, intelligent monitoring and artificial intelligence research. Modern unmanned aerial vehicles play an increasingly important role and are currently valued by all parties. With the higher requirements of people on intelligence, unmanned aerial vehicles are naturally favored by various industries: unmanned aerial vehicle recording at concert scene, unmanned aerial vehicle delivery of in order abundant express delivery, unmanned aerial vehicle shooting of outdoor exploration, etc. show that unmanned aerial vehicle has been applied to in people's daily life betterly, bring a great deal of facility for people. In recent years, the realization of real-time monitoring of unmanned aerial vehicles has shown great military and civil values, and has attracted great importance in academia and industry, as a typical video-based moving object detection and tracking problem, how to apply the prior art to the video monitoring of unmanned aerial vehicle moving objects, realize the real-time detection and tracking of unmanned aerial vehicle targets, and the technology has remarkable economic and social benefits in many aspects such as military guard, public security and the like.
Because the small unmanned aerial vehicle target has the characteristics of small size, variable flying speed, complex flying environment and the like, the method of radar detection, passive positioning and the like is only used to be easily influenced by other signal clutter, false alarm results are generated more, and the obtained results are possibly only a few pixels, only the position information of the unmanned aerial vehicle target is obtained, the flying area and flying machine of the unmanned aerial vehicle cannot be monitored with high precision, and accurate target positioning cannot be provided for subsequent interference interception, so that ideal results are difficult to have. In recent years, unmanned aerial vehicle identification and tracking methods based on optical image processing appear, but the effect is not satisfactory.
Through searching, chinese patent application CN201911268966.2, CN110706266A discloses an air target tracking method based on YOLOv3, comprising the following steps: generating a model file; video files are collected in real time, and two threads of YOLOv3 target detection and KCF target tracking are created; performing target detection by a YOLOv3 target tracking thread; transmitting the target position information in the step S03 to a KCF target tracking thread, and simultaneously executing the step S07 and the step S11; starting a KCF target tracking thread, and judging whether the KCF target tracking thread is initialized; manually setting a detection frame; finishing KCF parameter initialization; performing target detection by a KCF target tracking thread; taking a detection frame with the largest response value as a target; updating the position parameters; and obtaining final target position information. Although YOLOv3 is used in this patent, the speed of tracking is still further improved.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a video-based unmanned aerial vehicle identification and tracking method and device, which greatly improve the real-time performance of tracking.
In order to solve the technical problems, the invention is realized by the following technical scheme:
according to a first aspect of the present invention, there is provided a video-based unmanned aerial vehicle identification and tracking method, comprising:
s11, obtaining unmanned aerial vehicle labeling image samples with a plurality of models and different sizes as a data set;
s12, training the data set by using a YOLOv3 network to obtain a trained deep learning target detection model;
s13, improving the image quality of an input video by adopting a Retinex image enhancement method, identifying each frame of the input unmanned aerial vehicle video through the trained YOLOv3 deep learning target detection model, and obtaining a target unmanned aerial vehicle detection frame of each frame so as to prepare for a follow-up tracking task;
and S14, according to the identification result of the S13, adopting a Sort algorithm to realize rapid tracking of the unmanned aerial vehicle in the video.
The invention adopts the improvement based on the YOLOv3 network and the Sort tracking algorithm, and can ensure good precision while improving the tracking speed.
Preferably, the step S11 specifically includes:
collect a large amount of images that contain unmanned aerial vehicle, cover unmanned aerial vehicle of various models, every unmanned aerial vehicle model number many images set up unmanned aerial vehicle image and be unified size, carry out unmanned aerial vehicle's mark one by one to every image.
Preferably, in S12, the YOLOv3 network trains the data set, adjusts the network super-parameters, and obtains a deep learning target detection model with stable gradient descent, expected value reduction of the loss function and required fitting degree.
Preferably, an attention mechanism is added into the dark-53 of the YOLOv3 network to extract important features of data quickly and improve network identification effect, the attention mechanism can focus attention on important information to save system resources, a common convolutional neural network pooling layer is too simple and rough directly in a mode of maximum pooling or average pooling and key information cannot be identified, and therefore the attention mechanism can be adopted to improve the problem and improve the accuracy of a model.
Preferably, the YOLOv3 network, wherein the loss function uses a GIoU function as an index for measuring target detection positioning performance:
in the above formula, A represents a prediction frame, B represents a real frame, C represents the minimum closed area including A and B, and the molecule represents the area of the area in C which is not covered by A and B at the same time; the loss function value GIoU ranges from-1 to 1, so that the relation between the prediction frame and the real frame can be better reflected, and IoU is IoU loss function value in YOLOv3 network. The improved loss function IoU is GIoU, so that the relation between the predicted frame and the real frame can be better embodied, and the network identification accuracy can be improved. .
Preferably, in S13, the method further includes:
the image of the input video is converted into a constant image, so that the high fidelity of the image and the compression of the dynamic range of the image can be maintained, the color is enhanced, the color constancy is maintained, and the subsequent robustness of the identification network is improved.
The constancy image r (x, y) is
In the above formula, K is the number of Gaussian center surrounding functions of 1,2,3 and w respectively k Is the corresponding one of the kth scaleWeight of F k (x, y) kth center-surround function.
Preferably, in S14, the implementing fast tracking of the unmanned aerial vehicle in the video by using the Sort algorithm includes:
in each frame, taking the detected unmanned aerial vehicle detection frame as a reference, simultaneously adopting a Kalman filter to predict a tracking frame of the unmanned aerial vehicle, calculating IoU between all target detection frames of the current frame and all tracking frames predicted by Kalman, obtaining an optimal matching pair of the detection frames and the tracking frames IoU through a Hungary algorithm, representing the matched detection frames as a tracking result of the current frame, updating the Kalman tracker by using the currently detected target position information, and continuously matching the prediction frames of the next frame with the detection frames of the next frame;
and repeating the process to realize continuous tracking of the unmanned aerial vehicle.
According to a second aspect of the present invention there is provided a video-based drone identification and tracking device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor being operable to perform the video-based drone identification and tracking method when the program is executed.
Compared with the prior art, the invention has the following beneficial effects:
according to the unmanned aerial vehicle identification and tracking method based on the video, provided by the invention, the network model is trained according to a large number of data sets, unmanned aerial vehicle identification and tracking is performed by using a deep learning method, the existing network is improved, the image is enhanced, and the identification and tracking result which is more accurate and has stronger effect robustness is obtained.
The method for identifying and tracking the unmanned aerial vehicle based on the video provided by the invention is very fast on the premise of ensuring higher tracking precision and can be suitable for actual target tracking tasks under the condition that the real-time performance and the precision of the current target tracking cannot be considered.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:
FIG. 1 is a flow chart of a video-based drone identification and tracking method according to one embodiment of the present invention;
FIG. 2 is a schematic diagram of a network architecture modified by Darknet-53 of a Yolov 3-based network according to an embodiment of the present invention;
fig. 3 is a flow chart of a video-based unmanned aerial vehicle identification and tracking method according to an embodiment of the present invention.
Detailed Description
The following describes embodiments of the present invention in detail: the embodiment is implemented on the premise of the technical scheme of the invention, and detailed implementation modes and specific operation processes are given. It should be noted that variations and modifications can be made by those skilled in the art without departing from the spirit of the invention, which falls within the scope of the invention.
Fig. 1 is a flowchart of a video-based unmanned aerial vehicle identification and tracking method according to an embodiment of the present invention.
Referring to fig. 1, the method for identifying and tracking a video-based unmanned aerial vehicle according to the present embodiment includes the following steps:
s11, obtaining unmanned aerial vehicle labeling image samples of a plurality of models and different sizes;
collecting a large number of images containing unmanned aerial vehicles, covering unmanned aerial vehicles of various models, carrying out unmanned aerial vehicle annotation on each image one by each unmanned aerial vehicle model number, and obtaining unmanned aerial vehicle annotation image samples as a training data set.
And S12, training the data set obtained in S11 on the basis of the YOLOv3 network to obtain a trained deep learning target detection model.
S13, improving the image quality of the unmanned aerial vehicle video to be detected by adopting a Retinex image enhancement means, identifying each frame of the unmanned aerial vehicle video to be detected through a deep learning target detection model, and obtaining a target unmanned aerial vehicle detection frame of each frame so as to prepare for a follow-up tracking task;
s14, according to the detection frame of the target unmanned aerial vehicle obtained in the S13, a Sort algorithm is adopted to realize rapid tracking of the unmanned aerial vehicle in the video.
According to the method, the network model is trained according to a large number of data sets, unmanned aerial vehicle identification tracking is performed by using a deep learning method, accuracy and robustness of unmanned aerial vehicle identification and tracking are improved, and when an unmanned aerial vehicle image is unclear, image enhancement can be performed, so that the method is suitable for various complex scenes. While the tracking speed is improved, good precision can be ensured.
In a preferred embodiment, the above S11 may use 2664 unmanned aerial vehicle pictures as the training data set, where the pictures basically cover unmanned aerial vehicles in various types, different states, and various backgrounds, and the sizes of the images are the same. Of course, the number of pictures is merely illustrative, and in other embodiments, other numbers of unmanned pictures are possible, and not limited to 2664.
In another preferred embodiment, the step S12 uses the data set obtained in the training step S11 based on the YOLOv3 network to adjust the network super-parameters, so as to obtain a deep learning model with stable gradient descent, expected value of the loss function and required fitting degree. The embodiment applies the YOLOv3 network which is commonly applied to the image fields of vehicles and the like to unmanned aerial vehicle identification and tracking. In order to obtain better effects, the following improvements are made on the basis of the original YOLOv3 network:
1) The attention mechanism is added into the Darknet-53 of the YOLOv3 network, so that important characteristics of data can be extracted quickly, the network identification effect is improved, the attention mechanism can focus attention on important information, system resources are saved, a common convolutional neural network pooling layer is too simple and rough in a mode of maximum pooling or average pooling directly, key information cannot be identified, and therefore the problem can be improved by adopting the attention mechanism, and the accuracy of a model is improved.
2) The improved loss function IoU (Intersection over Union) is GIoU (Generalized Intersection over Union), so that the relation between the predicted frame and the real frame can be better embodied, and the defect of IoU is overcome.
IoU is adopted in the YOLOv3 network as an index for measuring the target detection positioning performance,
in the above formula, a represents a prediction frame, B represents a real frame, a numerator represents a union of the prediction frame and the real frame, and a denominator represents an intersection of the prediction frame and the real frame. However, if the predicted and real frames do not intersect, ioU is zero and cannot be optimized; even IoU, which is the same, cannot represent the same detection effect. GIoU has made improvements to the above problems,
in the above formula, C represents the minimum closed area including A and B, and the molecule represents the area of C which is not covered by A and B at the same time. As IoU ranges from 0 to 1 and GIoU ranges from-1 to 1, the relation between the prediction frame and the real frame can be better reflected, and the network identification accuracy can be improved.
As shown in fig. 3, the network structure modified by dark-53 based on YOLOv3 network specifically includes 52-layer convolution and 23 residual units. The network is a full convolution network, largely using residual layer jump connections, and in order to reduce the gradient negative effects of pooling, the pooling layer is abandoned and the convolution with step size 2 is used for 5 downsampling. In the 5-stage downsampling process, the convolution layer is followed by the residual unit and the attention mechanism. For example, if the input is 416x416, the output is 13x13 (416/2 5 =13), thereby performing a size conversion of the tensor. Through the improvements, the accuracy and the robustness of unmanned aerial vehicle identification and tracking can be well improved.
In another preferred embodiment, the step S13 may convert the image of the unmanned aerial vehicle video into a constant image, and may enhance the color while maintaining the high fidelity of the image and compressing the dynamic range of the image, thereby maintaining the color constant. Specifically, the constancy image r (x, y) is
In the above formula, K is 3 and represents RGB three channels, w k Is the weight corresponding to the kth scale, the values are all 1/3, the three scales are 15, 101, 301, S (x, y) is the observed image, F k (x, y) kth center-surround function.
In another preferred embodiment, S14, the rapid tracking of the unmanned aerial vehicle in the video by using the Sort algorithm may be implemented according to the following method:
and detecting each frame of the input unmanned aerial vehicle video through the trained YOLOv3 deep learning target detection model to obtain a target unmanned aerial vehicle detection frame of each frame. In each frame, taking the detected unmanned aerial vehicle detection frame as a reference, simultaneously adopting a Kalman filter to predict the unmanned aerial vehicle tracking frame, calculating IoU between all target detection frames of the current frame and all Kalman predicted tracking frames, obtaining the optimal matching pair of the detection frames and the tracking frames IoU through a Hungary algorithm, representing the matched detection frames as the tracking result of the current frame, updating the Kalman tracker by using the currently detected target position information, and continuing to match the predicted frames of the next frame with the detection frames of the next frame. The continuous tracking of the target can be realized.
According to the unmanned aerial vehicle identification and tracking method based on the video, the network model is trained according to a large number of data sets, unmanned aerial vehicle identification and tracking is performed by using a deep learning method, an existing network is improved, images are enhanced, and an identification tracking result which is more accurate and has stronger effect robustness is obtained.
In another embodiment, the invention also provides a video-based unmanned aerial vehicle identifying and tracking device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor can be used for executing the video-based unmanned aerial vehicle identifying and tracking method of the above embodiments when executing the program.
Optionally, a memory for storing a program; memory, which may include volatile memory (english) such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random-access memory (Double Data Rate Synchronous Dynamic Random Access Memory, DDR SDRAM), and the like; the memory may also include a non-volatile memory (English) such as a flash memory (English). The memory 62 is used to store computer programs (e.g., application programs, functional modules, etc. that implement the methods described above), computer instructions, etc., which may be stored in one or more memories in a partitioned manner. And the above-described computer programs, computer instructions, data, etc. may be invoked by a processor.
The computer programs, computer instructions, etc. described above may be stored in one or more memories in partitions. And the above-described computer programs, computer instructions, data, etc. may be invoked by a processor.
A processor for executing the computer program stored in the memory to implement the steps in the method according to the above embodiment. Reference may be made in particular to the description of the embodiments of the method described above.
The processor and the memory may be separate structures or may be integrated structures that are integrated together. When the processor and the memory are separate structures, the memory and the processor may be connected by a bus coupling.
Embodiments of the present application also provide a computer-readable storage medium having stored therein computer-executable instructions that, when executed by at least one processor of a user device, perform the various possible methods described above.
The embodiments disclosed herein were chosen and described in detail in order to best explain the principles of the invention and the practical application, and to thereby not limit the invention. Any modifications or variations within the scope of the description that would be apparent to a person skilled in the art are intended to be included within the scope of the invention. The above preferred features may be used alone in any of the embodiments, or in any combination without interfering with each other.
Claims (3)
1. The unmanned aerial vehicle identification and tracking method based on the video is characterized by comprising the following steps of:
s11, obtaining unmanned aerial vehicle labeling image samples with a plurality of models and different sizes as a data set;
s12, training the data set by using a YOLOv3 network to obtain a trained deep learning target detection model;
s13, improving the image quality of an input video by adopting a Retinex image enhancement method, and identifying each frame of the input unmanned aerial vehicle video through the trained YOLOv3 deep learning target detection model to obtain a target unmanned aerial vehicle detection frame of each frame;
s14, according to the identification result of the S13, adopting a Sort algorithm to realize rapid tracking of the unmanned aerial vehicle in the video;
adding an attention mechanism into a Darknet-53 of the Yolov3 network to rapidly extract important features of data;
in S14, the implementing fast tracking of the unmanned aerial vehicle in the video by using the Sort algorithm includes:
in each frame, taking the detected unmanned aerial vehicle detection frame as a reference, simultaneously adopting a Kalman filter to predict a tracking frame of the unmanned aerial vehicle, calculating IoU between all target detection frames of the current frame and all tracking frames predicted by Kalman, obtaining an optimal matching pair of the detection frames and the tracking frames IoU through a Hungary algorithm, representing the matched detection frames as a tracking result of the current frame, updating the Kalman tracker by using the currently detected target position information, and continuously matching the prediction frames of the next frame with the detection frames of the next frame;
repeating the above process to realize continuous tracking of the unmanned aerial vehicle;
s12, training the data set by the YOLOv3 network, and adjusting network super-parameters to obtain a deep learning target detection model with stable gradient descent, expected value of loss function and required fitting degree;
the YOLOv3 network, wherein the loss function adopts a GIoU function as an index for measuring target detection positioning performance:
in the above formula, A represents a prediction frame, B represents a real frame, C represents the minimum closed area including A and B, and the molecule represents the area of the area in C which is not covered by A and B at the same time; the range of the loss function value GIoU is from-1 to 1, the relation between the prediction frame and the real frame can be better reflected, and IoU is IoU loss function value in the YOLOv3 network;
in S13, further including:
converting an image of an input video into a constant image, the constant image r (x, y) being:
in the above formula, K is the number of Gaussian center surrounding functions of 1,2,3 and w respectively k Is the weight corresponding to the kth scale, S (x, y) is each frame of image of the observed image input video, x, y represents a specific position in the image, F k (x, y) kth center-surround function.
2. The video-based unmanned aerial vehicle identification and tracking method of claim 1, wherein S11 is specifically: collect a large amount of images that contain unmanned aerial vehicle, cover unmanned aerial vehicle of various models, every unmanned aerial vehicle model number many images set up unmanned aerial vehicle image and be unified size, carry out unmanned aerial vehicle's mark one by one to every image.
3. A video-based drone identification and tracking device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor is operable to perform the method of any one of claims 1-2 when the program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010231230.4A CN111460968B (en) | 2020-03-27 | 2020-03-27 | Unmanned aerial vehicle identification and tracking method and device based on video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010231230.4A CN111460968B (en) | 2020-03-27 | 2020-03-27 | Unmanned aerial vehicle identification and tracking method and device based on video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111460968A CN111460968A (en) | 2020-07-28 |
| CN111460968B true CN111460968B (en) | 2024-02-06 |
Family
ID=71680515
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010231230.4A Active CN111460968B (en) | 2020-03-27 | 2020-03-27 | Unmanned aerial vehicle identification and tracking method and device based on video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111460968B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111931654A (en) * | 2020-08-11 | 2020-11-13 | 精英数智科技股份有限公司 | Intelligent monitoring method, system and device for personnel tracking |
| CN112184767A (en) * | 2020-09-22 | 2021-01-05 | 深研人工智能技术(深圳)有限公司 | Method, device, equipment and storage medium for tracking moving object track |
| CN112132032B (en) * | 2020-09-23 | 2024-07-12 | 平安国际智慧城市科技股份有限公司 | Traffic sign board detection method and device, electronic equipment and storage medium |
| CN112288770A (en) * | 2020-09-25 | 2021-01-29 | 航天科工深圳（集团）有限公司 | Video real-time multi-target detection and tracking method and device based on deep learning |
| CN112348057A (en) * | 2020-10-20 | 2021-02-09 | 歌尔股份有限公司 | Target identification method and device based on YOLO network |
| CN112419368A (en) * | 2020-12-03 | 2021-02-26 | 腾讯科技（深圳）有限公司 | Method, device and equipment for tracking track of moving target and storage medium |
| CN112465854A (en) * | 2020-12-17 | 2021-03-09 | 北京三川未维科技有限公司 | Unmanned aerial vehicle tracking method based on anchor-free detection algorithm |
| CN113139419B (en) * | 2020-12-28 | 2024-05-31 | 西安天和防务技术股份有限公司 | Unmanned aerial vehicle detection method and device |
| CN112734794B (en) * | 2021-01-14 | 2022-12-23 | 北京航空航天大学 | Moving target tracking and positioning method based on deep learning |
| CN112819858B (en) * | 2021-01-29 | 2024-03-22 | 北京博雅慧视智能技术研究院有限公司 | Target tracking method, device, equipment and storage medium based on video enhancement |
| CN112906523B (en) * | 2021-02-04 | 2022-12-27 | 上海航天控制技术研究所 | Hardware-accelerated deep learning target machine type identification method |
| CN112884811A (en) * | 2021-03-18 | 2021-06-01 | 中国人民解放军国防科技大学 | Photoelectric detection tracking method and system for unmanned aerial vehicle cluster |
| CN114140501A (en) * | 2022-01-30 | 2022-03-04 | 南昌工程学院 | Target tracking method and device and readable storage medium |
| CN117455955B (en) * | 2023-12-14 | 2024-03-08 | 武汉纺织大学 | Pedestrian multi-target tracking method based on unmanned aerial vehicle visual angle |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101674490A (en) * | 2009-09-23 | 2010-03-17 | 电子科技大学 | Color image color constant method based on retina vision mechanism |
| CN101706964A (en) * | 2009-08-27 | 2010-05-12 | 北京交通大学 | Color constancy calculating method and system based on derivative structure of image |
| WO2018133666A1 (en) * | 2017-01-17 | 2018-07-26 | 腾讯科技（深圳）有限公司 | Method and apparatus for tracking video target |
| CN110070561A (en) * | 2019-03-13 | 2019-07-30 | 合肥师范学院 | A kind of image enhancement of blurred vision condition and method for tracking target and system |
| CN110516556A (en) * | 2019-07-31 | 2019-11-29 | 平安科技（深圳）有限公司 | Multi-target tracking detection method, device and storage medium based on Darkflow-DeepSort |
| CN110706266A (en) * | 2019-12-11 | 2020-01-17 | 北京中星时代科技有限公司 | Aerial target tracking method based on YOLOv3 |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019231217A1 (en) * | 2018-05-28 | 2019-12-05 | Samsung Electronics Co., Ltd. | Method and system for dnn based imaging |
-
2020
- 2020-03-27 CN CN202010231230.4A patent/CN111460968B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101706964A (en) * | 2009-08-27 | 2010-05-12 | 北京交通大学 | Color constancy calculating method and system based on derivative structure of image |
| CN101674490A (en) * | 2009-09-23 | 2010-03-17 | 电子科技大学 | Color image color constant method based on retina vision mechanism |
| WO2018133666A1 (en) * | 2017-01-17 | 2018-07-26 | 腾讯科技（深圳）有限公司 | Method and apparatus for tracking video target |
| CN110070561A (en) * | 2019-03-13 | 2019-07-30 | 合肥师范学院 | A kind of image enhancement of blurred vision condition and method for tracking target and system |
| CN110516556A (en) * | 2019-07-31 | 2019-11-29 | 平安科技（深圳）有限公司 | Multi-target tracking detection method, device and storage medium based on Darkflow-DeepSort |
| CN110706266A (en) * | 2019-12-11 | 2020-01-17 | 北京中星时代科技有限公司 | Aerial target tracking method based on YOLOv3 |
Non-Patent Citations (1)
| Title |
|---|
| 徐义鎏，贺鹏.改进损失函数的Yolov3车型检测算法.《信息通信》.2019,(第12期),4-7. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111460968A (en) | 2020-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111460968B (en) | Unmanned aerial vehicle identification and tracking method and device based on video | |
| CN110188705B (en) | Remote traffic sign detection and identification method suitable for vehicle-mounted system | |
| CN107153817B (en) | Pedestrian re-identification data labeling method and device | |
| Cepni et al. | Vehicle detection using different deep learning algorithms from image sequence | |
| CN113361645B (en) | Target detection model construction method and system based on meta learning and knowledge memory | |
| CN111640138B (en) | Target tracking method, device, equipment and storage medium | |
| CN110992378B (en) | Dynamic updating vision tracking aerial photographing method and system based on rotor flying robot | |
| CN109919223B (en) | Target detection method and device based on deep neural network | |
| CN112200031A (en) | Network model training method and equipment for generating image corresponding word description | |
| Feng et al. | A novel saliency detection method for wild animal monitoring images with WMSN | |
| CN114898290A (en) | Real-time detection method and system for marine ship | |
| CN111079543A (en) | Efficient vehicle color identification method based on deep learning | |
| CN112668662B (en) | Outdoor mountain forest environment target detection method based on improved YOLOv3 network | |
| CN112712066B (en) | Image recognition method and device, computer equipment and storage medium | |
| Chen et al. | Object counting in remote sensing via selective spatial‐frequency pyramid network | |
| Li et al. | Incremental learning of infrared vehicle detection method based on SSD | |
| CN117152790A (en) | Method and system for detecting cow face in complex scene | |
| CN116704187A (en) | Real-time semantic segmentation method, system and storage medium for semantic alignment | |
| Hong et al. | Improved SSD model for pedestrian detection in natural scene | |
| US20240221426A1 (en) | Behavior detection method, electronic device, and computer readable storage medium | |
| CN113780462B (en) | Vehicle detection network establishment method based on unmanned aerial vehicle aerial image and application thereof | |
| CN111582107B (en) | Training method and recognition method of target re-recognition model, electronic equipment and device | |
| CN114694042A (en) | Disguised person target detection method based on improved Scaled-YOLOv4 | |
| CN111079704A (en) | Face recognition method and device based on quantum computation | |
| Yuan et al. | Vehicle detection based on area and proportion prior with faster-RCNN |
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 |