CN111931554A - Target detection method and model for ship emitting black smoke - Google Patents
Target detection method and model for ship emitting black smoke Download PDFInfo
- Publication number
- CN111931554A CN111931554A CN202010535227.1A CN202010535227A CN111931554A CN 111931554 A CN111931554 A CN 111931554A CN 202010535227 A CN202010535227 A CN 202010535227A CN 111931554 A CN111931554 A CN 111931554A
- Authority
- CN
- China
- Prior art keywords
- target
- picture
- model
- ship
- black smoke
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 33
- 239000000779 smoke Substances 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 24
- 238000010586 diagram Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 abstract description 3
- 238000004590 computer program Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000011895 specific detection Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
-
- 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
- G06V2201/08—Detecting or categorising vehicles
Abstract
The invention discloses a target detection method and a model of a ship emitting black smoke, and the method comprises the following steps: inputting original pictures in a picture set into a front end of a model for detection to obtain a target area, and obtaining a target picture according to the target area; and step two, inputting the target picture into the rear end of the model for training to finally obtain the target recognition model. According to the method, before the ship emitting black smoke is identified, the ship area needing to be identified is obtained at the front end of the model, then the target detection is carried out on the black smoke around the ship, and finally the technical effect of improving the target detection accuracy can be achieved.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to a target detection method and a model of a ship emitting black smoke.
Background
Due to the fact that the fuel injector of the marine main engine emits a large amount of black smoke in the combustion process due to poor conditions, on one hand, serious pollution is caused to the atmosphere; on the other hand, potential safety hazards exist. There is therefore a need to develop a method of identifying black-smoking vessels that provides technical support for law enforcement and related management personnel to make rapid decisions.
The target detection is to find out all interested objects in the image, and includes two subtasks of object positioning and object classification, however, most of the existing target detection is to detect a fixed object, the appearance, shape or posture of the existing target detection is static, and the existing target detection method cannot be applied to ship identification which emits black smoke or has poor identification effect due to the complex and diverse appearance, shape, posture and environment background (multiple sea surface environments) of black smoke.
Therefore, it is necessary to research a target detection technology suitable for a ship which emits black smoke and has high real-time performance and high precision.
Disclosure of Invention
The invention provides a target detection method of a black smoke-emitting ship with higher real-time performance and precision.
The invention is realized by the following technical scheme:
a target detection method for a ship emitting black smoke, comprising the following steps:
step one, inputting original pictures in a picture set into the front end of a model to be detected, obtaining a target area, and obtaining a target picture according to the target area;
and step two, inputting the target picture into the rear end of the model to train the network, and finally obtaining the target recognition model.
Preferably, the first step of the present invention specifically comprises:
step S11, extracting the characteristics of the input picture through a convolutional neural network at the front end of the model, and generating default boxes corresponding to the characteristic graphs by each detector;
step S12, matching each default box with a group channel box to ensure that each group channel box can correspond to a plurality of default boxes;
and step S13, utilizing the maximum value to restrain and screen all the generated default boxes to obtain the final target area.
Preferably, the picture set constructing method of the first step of the present invention is: the method comprises the steps of collecting videos including black smoke ship scenes and videos of a series of normal ship scenes, processing the videos of all the scenes in a classified mode, cutting the videos to obtain a frame of picture, forming a picture set, and using the picture set as input of the front end of a model.
Preferably, the second step of the present invention specifically comprises:
step S21, converting the target picture into an input dimension required by an MBConv module through a Conv3 multiplied by 3 layer;
step S22, inputting the target picture converted in the step S21 into a plurality of MBConv modules, extracting feature maps, and adjusting the parameters of each MBConv module;
step S23, a feature diagram self-adaptive connection method based on Fully-conditional-Neural-Network is adopted, so that the Conv1x1 Network can adapt to feature diagrams of various sizes and is unified into required dimensions;
and step S24, finishing the classification and identification of the pictures through the output feature map, and training to obtain a final classifier.
Preferably, the method further comprises a third step of identifying and classifying the picture to be detected by using the trained target identification model.
On the other hand, the invention also provides a target detection model of the ship emitting black smoke, which comprises a front-end detector and a rear-end training module;
the front-end detector is used for detecting an input original picture to obtain a target area, obtaining a target picture according to the target area and transmitting the target picture to the rear-end training module;
and the back-end training module trains by using the input target picture to obtain a target detection model.
Preferably, the front-end detector of the present invention is configured to perform the following operations:
A. extracting the characteristics of an input picture through a convolutional neural network at the front end of the model, and generating default boxes corresponding to characteristic graphs by each detector;
B. matching each default box with a group channel box to ensure that each group channel box can correspond to a plurality of default boxes;
C. utilizing a maximum value to inhibit and screen all generated default boxes to obtain a final target area;
D. and acquiring a target picture from the input picture according to the target area.
Preferably, the back-end training module of the present invention is configured to perform the following operations:
a1, converting the target picture into an input dimension required by an MBConv module through a Conv3 multiplied by 3 layer;
a2, inputting the target picture converted by A1 into a plurality of MBConv modules, extracting feature maps, and adjusting the parameters of each MBConv module;
a3, adopting a feature diagram self-adaptive connection method based on Fully-conditional-Neural-Network to enable a Conv1x1 Network to adapt to feature diagrams of various sizes and unify the feature diagrams into required dimensions;
and A4, finishing the classification and identification of the pictures through the output feature diagram, and training to obtain a final classifier.
Preferably, the original picture input by the model of the present invention is: the method comprises the steps of collecting videos including black smoke ship scenes and videos of a series of normal ship scenes, processing the videos of all the scenes in a classified mode, and cutting the videos to obtain one frame of picture.
The invention has the following advantages and beneficial effects:
according to the method, before the ship emitting black smoke is identified, the ship area needing to be identified is obtained at the front end of the model, then the target detection is carried out on the black smoke around the ship, and finally the technical effect of improving the target detection accuracy can be achieved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:
FIG. 1 is a schematic flow chart of the method of the present invention.
FIG. 2 is a schematic diagram of the model structure of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.
Example 1
The embodiment provides a target detection method for a ship emitting black smoke.
As shown in fig. 1, the target detection method of the present invention mainly includes the following steps:
firstly, an original data set is constructed, original pictures in the data set are input into a front-end detector of a model to be detected, target pictures are obtained, and the target pictures form a training data set.
In this embodiment, the method for constructing the original data set includes: the camera is used for collecting videos of a certain area, the videos comprise illegal scenes of ships emitting black smoke and a series of normal ship scenes, and the videos of all the scenes are sorted. And then cutting the video to obtain a data set formed by one frame of pictures, and taking the data set as the input of the neural network.
In this embodiment, the model front-end detector adopts an SSD detector, and the specific detection process is as follows:
(1) in the network, extracting picture features by using a convolutional neural network, and generating default boxes by using each detector corresponding to a feature map;
(2) matching default boxes, matching each default box with a group channel, ensuring that each group channel can correspond to a plurality of default boxes, and avoiding missing detection;
(3) measuring default boxes, and calculating the value of softmax loss or cross entry loss of the default box belonging to the background category when the ith default box is matched with the jth group of the jth group;
wherein, ViIs the output of the preceding output unit of the classifier, i represents the class index, C is the total number of classes, SiRepresenting the ratio of the index of the current element to the sum of the indices of all elements.
(4) All the generated default boxes are screened by NMS (maximum suppression) to obtain a final target area;
wherein N istIs a threshold value of NMS, biIs the detected bounding box, siIs corresponding to biConfidence of the box, M is the union of all detected boxes.
(5) And determining the position of the target in the original picture according to the obtained target area, namely obtaining an input picture for identifying the ship black smoke network.
And secondly, training a training module at the rear end of the model by using the target picture to obtain a target recognition model.
In this embodiment, the training module at the back end of the model adopts the MBConv module, and the specific training process is as follows:
(1) converting the picture into an input dimension required by an MBConv module through a first Conv3x3 layer;
(2) the feature map is extracted from the picture through a series of MBConv modules, the parameters of each MBConv module are finely adjusted to adapt to the current use environment, and the network can obtain a better receptive field by a combined scale optimization method;
depth:d＝αφ
width:ω＝βφ
resolution:r＝γφ
wherein, α. β2·γ2Is approximately equal to 2, alpha is more than or equal to 1, beta is more than or equal to 1, gamma is more than or equal to 1, and phi is a composite coefficient.
(3) And (3) utilizing a feature diagram self-adaptive connection mode based on the Fully-conditional-Neural-Network. The Conv1x1 network can be adapted to feature maps of various sizes, outputs 1280 are obtained after fc7 layers, outputs 2 are obtained after fc8 layers, and the outputs are unified into required dimensionalities, namely black smoke emission and black smoke non-emission;
(4) and finally, finishing the classification, identification and detection of the pictures through the output characteristic graph to obtain the required target identification model.
And thirdly, detecting and identifying the picture to be detected by utilizing the trained target identification model.
Experiments prove that one frame of picture in the video stream is intercepted at intervals of 1s, the system can monitor illegal ships which smoke in real time, and the recognition accuracy is good.
Example 2
The embodiment provides a target detection model of a ship emitting black smoke.
In the embodiment, video information of two types of scenes (black smoke and normal ship scenes) is collected, the video is cut into one frame and one frame of pictures to be input into a network, the regions where the targets are located are obtained through detection of the front end of the model, then the regions are cut out from the pictures, and a series of cut-out pictures are input into the rear end of the model to be trained to finally obtain the target detection model.
Specifically, as shown in fig. 2, the detection model of the present embodiment includes:
a model front-end detector and a model back-end training module.
The model front-end detector of the embodiment is used for detecting an input original picture to obtain a target area, obtaining a target picture according to the target area and transmitting the target picture to the rear-end training module;
in this embodiment, the target image is input into the model back-end training module for training, so as to obtain the target detection model.
The model front-end detector of the present embodiment is configured to implement the following operations:
A. extracting the characteristics of an input picture through a convolutional neural network at the front end of the model, and generating default boxes corresponding to characteristic graphs by each detector;
B. matching each default box with a group channel box to ensure that each group channel box can correspond to a plurality of default boxes;
C. utilizing a maximum value to inhibit and screen all generated default boxes to obtain a final target area;
D. and acquiring a target picture from the input picture according to the target area.
The model back-end training module of the present embodiment is configured to perform the following operations:
a1, converting the target picture into an input dimension required by an MBConv module through a Conv3 multiplied by 3 layer;
a2, inputting the target picture converted by A1 into a plurality of MBConv modules, extracting feature maps, and adjusting the parameters of each MBConv module;
a3, adopting a feature diagram self-adaptive connection method based on Fully-conditional-Neural-Network to enable a Conv1x1 Network to adapt to feature diagrams of various sizes and unify the feature diagrams into required dimensions;
and A4, finishing the classification and identification of the pictures through the output feature diagram, and training to obtain a final classifier.
The embodiment utilizes the detector at the front end of the model to extract the black smoke target detection area, reduces the influence of the background on the identification, and is favorable for improving the precision. Experiments prove that the model accuracy of the invention is higher than that of the model using EfficientNet alone, and the real-time performance is better.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (9)
1. A target detection method for a ship emitting black smoke is characterized by comprising the following steps:
step one, inputting original pictures in a picture set into the front end of a model to be detected, obtaining a target area, and obtaining a target picture according to the target area;
and step two, inputting the target picture into the rear end of the model to train the network, and finally obtaining the target recognition model.
2. The method for detecting the target of the ship which emits the black smoke according to claim 1, wherein the first step specifically comprises the following steps:
step S11, extracting the characteristics of the input picture through a convolutional neural network at the front end of the model, and generating default boxes corresponding to the characteristic graphs by each detector;
step S12, matching each default box with a group channel box to ensure that each group channel box can correspond to a plurality of default boxes;
and step S13, utilizing the maximum value to restrain and screen all the generated default boxes to obtain the final target area.
3. The method for detecting the target of the ship which emits the black smoke according to claim 1, wherein the picture set constructing method in the first step comprises the following steps: the method comprises the steps of collecting videos including black smoke ship scenes and videos of a series of normal ship scenes, processing the videos of all the scenes in a classified mode, cutting the videos to obtain a frame of picture, forming a picture set, and using the picture set as input of the front end of a model.
4. The method for detecting the target of the ship which emits the black smoke according to claim 1, wherein the second step specifically comprises the following steps:
step S21, converting the target picture into an input dimension required by an MBConv module through a Conv3 multiplied by 3 layer;
step S22, inputting the target picture converted in the step S21 into a plurality of MBConv modules, extracting feature maps, and adjusting the parameters of each MBConv module;
step S23, a feature diagram self-adaptive connection method based on Fully-conditional-Neural-Network is adopted, so that the Conv1x1 Network can adapt to feature diagrams of various sizes and is unified into required dimensions;
and step S24, finishing the classification and identification of the pictures through the output feature map, and training to obtain a final classifier.
5. The method for detecting the target of the ship emitting the black smoke according to any one of claims 1 to 4, further comprising a third step of identifying and classifying the picture to be detected by using a trained target identification model.
6. A target detection model of a ship emitting black smoke is characterized by comprising a front-end detector and a rear-end training module;
the front-end detector is used for detecting an input original picture to obtain a target area, obtaining a target picture according to the target area and transmitting the target picture to the rear-end training module;
and the back-end training module trains by using the input target picture to obtain a target detection model.
7. The black-smoking vessel target detection model of claim 6, wherein the front-end detector is configured to perform the following operations:
A. extracting the characteristics of an input picture through a convolutional neural network at the front end of the model, and generating default boxes corresponding to characteristic graphs by each detector;
B. matching each default box with a group channel box to ensure that each group channel box can correspond to a plurality of default boxes;
C. utilizing a maximum value to inhibit and screen all generated default boxes to obtain a final target area;
D. and acquiring a target picture from the input picture according to the target area.
8. The black-smoking vessel target detection model of claim 6, wherein the back-end training module is configured to perform the following operations:
a1, converting the target picture into an input dimension required by an MBConv module through a Conv3 multiplied by 3 layer;
a2, inputting the target picture converted by A1 into a plurality of MBConv modules, extracting feature maps, and adjusting the parameters of each MBConv module;
a3, adopting a feature diagram self-adaptive connection method based on Fully-conditional-Neural-Network to enable a Conv1x1 Network to adapt to feature diagrams of various sizes and unify the feature diagrams into required dimensions;
and A4, finishing the classification and identification of the pictures through the output feature diagram, and training to obtain a final classifier.
9. The model of claim 6, wherein the model inputs the following original pictures: the method comprises the steps of collecting videos including black smoke ship scenes and videos of a series of normal ship scenes, processing the videos of all the scenes in a classified mode, and cutting the videos to obtain one frame of picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010535227.1A CN111931554A (en) | 2020-06-12 | 2020-06-12 | Target detection method and model for ship emitting black smoke |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010535227.1A CN111931554A (en) | 2020-06-12 | 2020-06-12 | Target detection method and model for ship emitting black smoke |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111931554A true CN111931554A (en) | 2020-11-13 |
Family
ID=73317175
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010535227.1A Pending CN111931554A (en) | 2020-06-12 | 2020-06-12 | Target detection method and model for ship emitting black smoke |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111931554A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190102646A1 (en) * | 2017-10-02 | 2019-04-04 | Xnor.ai Inc. | Image based object detection |
| CN110569844A (en) * | 2019-08-26 | 2019-12-13 | 中国人民解放军91550部队 | ship recognition method and system based on deep learning |
| US20200143227A1 (en) * | 2018-11-06 | 2020-05-07 | Google Llc | Neural Architecture Search with Factorized Hierarchical Search Space |
-
2020
- 2020-06-12 CN CN202010535227.1A patent/CN111931554A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190102646A1 (en) * | 2017-10-02 | 2019-04-04 | Xnor.ai Inc. | Image based object detection |
| US20200143227A1 (en) * | 2018-11-06 | 2020-05-07 | Google Llc | Neural Architecture Search with Factorized Hierarchical Search Space |
| CN110569844A (en) * | 2019-08-26 | 2019-12-13 | 中国人民解放军91550部队 | ship recognition method and system based on deep learning |
Non-Patent Citations (4)
| Title |
|---|
| MINGXING TAN等: "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks", 《PROCEEDINGS OF THE 36 TH INTERNATIONAL CONFERENCE ON MACHINE LEARNING, LONG BEACH, CALIFORNIA, PMLR 97, 2019》 * |
| 姚红革等: "基于SSD的多特征刑侦场景识别", 《西安邮电大学学报》 * |
| 张朝阳: "《深入浅出 工业机器学习算法详解与实战》", 31 January 2020, 北京：机械工业出版社 * |
| 韩哲哲等: "基于火焰成像和堆栈降噪自编码的燃烧工况识别", 《东南大学学报(自然科学版)》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109977921B (en) | Method for detecting hidden danger of power transmission line | |
| CN109165585A (en) | A kind of improved ship target detection method based on YOLO V2 | |
| CN114266977B (en) | Multi-AUV underwater target identification method based on super-resolution selectable network | |
| CN110033481A (en) | Method and apparatus for carrying out image procossing | |
| CN113052211A (en) | Pruning method based on characteristic rank and channel importance | |
| CN114529799A (en) | Aircraft multi-target tracking method based on improved YOLOV5 algorithm | |
| CN111507371A (en) | Method and apparatus | |
| CN112613387A (en) | Traffic sign detection method based on YOLOv3 | |
| CN113963210A (en) | Deep learning-based detection method and sorting system for waste data storage equipment | |
| CN115512206A (en) | Improved YOLOv5 target detection method suitable for low-illumination environment | |
| CN114708518A (en) | Bolt defect detection method based on semi-supervised learning and priori knowledge embedding strategy | |
| CN112257612A (en) | Unmanned aerial vehicle video frame filtering method and device based on edge intelligence | |
| CN115115863A (en) | Water surface multi-scale target detection method, device and system and storage medium | |
| US11468572B2 (en) | Image processing device, image recognition device, image processing program, and image recognition program | |
| CN112668662B (en) | Outdoor mountain forest environment target detection method based on improved YOLOv3 network | |
| Zhang et al. | CE-RetinaNet: A channel enhancement method for infrared wildlife detection in UAV images | |
| CN110688918B (en) | Rapid behavior detection method based on long-term enhanced feature enhancement and sparse dynamic sampling | |
| CN111931554A (en) | Target detection method and model for ship emitting black smoke | |
| CN115953744A (en) | Vehicle identification tracking method based on deep learning | |
| CN110751065B (en) | Training data acquisition method and device | |
| KR20210031444A (en) | Method and Apparatus for Creating Labeling Model with Data Programming | |
| CN111563428A (en) | Airport parking space intrusion detection method and system | |
| CN113139077B (en) | Method, device, terminal and storage medium for identifying ship identity | |
| CN116343132B (en) | Complex scene power equipment defect identification method and device and computer equipment | |
| CN115908952B (en) | High-speed railway tunnel fixture detection method based on improved YOLOv5 algorithm |
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 | ||
| CB02 | Change of applicant information |
Address after: 610000 No. 270, floor 2, No. 8, Jinxiu street, Wuhou District, Chengdu, Sichuan Applicant after: Chengdu shuzhilian Technology Co.,Ltd. Address before: 610000 No.2, 4th floor, building 1, Jule Road intersection, West 1st section of 1st ring road, Wuhou District, Chengdu City, Sichuan Province Applicant before: CHENGDU SHUZHILIAN TECHNOLOGY Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201113 |
|
| RJ01 | Rejection of invention patent application after publication |