CN109165575A - A kind of pyrotechnics recognizer based on SSD frame - Google Patents
A kind of pyrotechnics recognizer based on SSD frame Download PDFInfo
- Publication number
- CN109165575A CN109165575A CN201810885395.6A CN201810885395A CN109165575A CN 109165575 A CN109165575 A CN 109165575A CN 201810885395 A CN201810885395 A CN 201810885395A CN 109165575 A CN109165575 A CN 109165575A
- Authority
- CN
- China
- Prior art keywords
- pyrotechnics
- layer
- detection model
- training
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 32
- 238000013500 data storage Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims description 9
- 238000005286 illumination Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 210000000352 storage cell Anatomy 0.000 claims description 3
- 238000000034 method Methods 0.000 abstract description 10
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 235000019504 cigarettes Nutrition 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- XEEYBQQBJWHFJM-UHFFFAOYSA-N iron Substances [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 239000000779 smoke Substances 0.000 description 1
- 238000013517 stratification Methods 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
-
- 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
- G06V20/44—Event 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)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Fire-Detection Mechanisms (AREA)
Abstract
It is complicated for high-speed rail monitor video background, noise is more, the problems such as picture easily exposes, study a kind of pyrotechnics recognizer based on picture depth study SSD frame, wherein detection model training network is attached most importance to the VGG16 network after structure, and the detection model training network after reconstruct increases 6 convolutional layers and 1 pond layer on the basis of VGG16.Realize the convolutional layer need the parameter designed include the number of filter, the size of filter, the initial method of parameter, biasing initial method, whether open bias term, every step-length for adding how many a pixels and filter on one side of input.A kind of pyrotechnics identifying system, including video acquisition unit, image enhancement processing unit, pyrotechnics recognition unit, data storage unit are devised using the detection model that this network training goes out.This system cooperates high-speed rail monitor video to use, even if the resolution ratio of image is lower, there are complex backgrounds in high-speed rail monitor video, also can guarantee the precision of pyrotechnics target detection.
Description
Technical field
The invention belongs to fire defector field, especially a kind of pyrotechnics recognizer based on SSD.
Background technique
In the high-iron carriage of high-speed cruising, once occur pyrotechnics alert, caused by loss will be inestimable, thus prevent
It is the most important thing with timely discovery pyrotechnics alert.Up to the present most widely used mainly temperature sensitive type fire detector and sense
Cigarette type fire detector.Temperature sensitive type fire detector and smoke-sensitive fire detector are the temperature and cigarette by perceiving flame periphery
Mistiness degree, information content and threshold value according to perception compare to determine whether alert.
As technology develops, there is flame based on image information, smog identifying system, this identifying system can be in real time
The image for handling acquisition, can greatly shorten pre-warning time, be advantageously implemented early prediction and the control of fire.Traditional algorithm
The identification of flame may be implemented, but recognition accuracy is low, rate of failing to report is higher.
Summary of the invention
High-speed rail monitor video is there are background complexity, the problems such as noise is more, and picture easily exposes, these complex environment factor meetings
The problem for causing fire defector precision low proposes a kind of pyrotechnics recognizer based on this, the technical solution adopted is as follows:
A kind of pyrotechnics recognizer based on SSD, detection model used in pyrotechnics recognizer is by the model after reconstructing
Network training obtains, and the prototype network is reconstructed VGG16 network, and the prototype network reduces one on the basis of VGG16
A full articulamentum remains two full articulamentums, increases 6 convolutional layers and 1 pond layer.
Further, the step of training detection model includes:
Step 1. pyrotechnics video acquisition；
Step 2. image preprocessing, the pyrotechnics video that specifically including will acquire are converted into the image sequence of one group of 10 frame, will
Pyrotechnics mark in image sequence is randomly assigned with the ratio of 1:3 with label at the image set for having label using classifier
Image set into test set and training set, the image in test set and training set is converted into LMDB format；
The pyrotechnics detection model that step 3. is reconstructed using training data training, until model is restrained；
Step 4. verifies pyrotechnics detection model using test data set；
Further, the 10th layer of convolutional layer, the 15th layer of convolution when training pyrotechnics detection model, in the prototype network
Layer, the 17th layer of convolutional layer, the 19th layer of convolutional layer, the 2nd layer of full articulamentum and the 6th pond layer are rolled up using two 3*3 convolution kernels
Product, obtains 6 groups of characteristic values.
Further, one in described two 3*3 convolution kernels whether there is for detecting target, another is for judging
Whether target is pyrotechnics.
Further, when training pyrotechnics detection model, the second full articulamentum is by the first full articulamentum in the prototype network
It is obtained after the convolution nuclear convolution of 1024 1*1, the first full articulamentum with the first full articulamentum by being connected in the prototype network
Pond layer obtained after the convolution nuclear convolution of 1024 3*3.
Further, the confidence level of the prototype network output is determined by 6 groups of characteristic value fusion calculations.
Compared with prior art, the beneficial effects of the present invention are: in high-speed rail video there are background complexity, noise
It is more, the problems such as picture easily exposes, 6 convolutional layers and 1 pond layer, the spy of this stratification are increased on the basis of VGG16
Sign extraction process can add up, and assign neural network powerful ability in feature extraction, trained model is made to obtain more complicated spy
Sign.The pond layer of addition make neural network ignore target inclination, rotation etc relative position variation, reduce characteristic pattern
Dimension, avoid over-fitting, improve the precision of model.
Another object of the present invention is to propose a kind of pyrotechnics identifying system, the pyrotechnics identifying system includes video acquisition
Unit, image enhancement processing unit, pyrotechnics recognition unit, data storage unit, wherein video acquisition unit is for obtaining high-speed rail
Monitor video, for handling the insufficient picture of illumination, pyrotechnics recognition unit is utilized by reconstruct image enhancement processing unit
The pyrotechnics detection model of detection model training network training is determined as pyrotechnics when the confidence level of pyrotechnics detection is greater than 0.5,
Data storage cell is used to store historical information when fire behavior occurs.
Further, the pyrotechnics identifying system further includes high-definition camera and warning device.
Compared with prior art, the beneficial effects of the present invention are above-mentioned pyrotechnics recognizer training is utilized in this system
Detection model, even if acquisition image resolution ratio it is lower, acquisition image in there are disturbing factors, also can guarantee pyrotechnics target
The precision of detection greatly improves alarm accuracy rate.
Detailed description of the invention
Fig. 1 is the detection model training network structure after present invention reconstruct；
Fig. 2 is the flow chart of trained detection model；
Fig. 3 is pyrotechnics detection system work flow diagram.
Specific embodiment
As shown in Figure 1, the detection model training network in the present invention is the prototype network based on VGG16 network reconfiguration, institute
State prototype network reduces a full articulamentum on the basis of VGG16, remains two full articulamentum fc6, fc7, increases
6 convolutional layers conv8_1, conv8_2, conv9_1, conv9_2, conv10_1, conv10_2 and 1 pond layer
pool11。
The foundation and training of detection model are realized based on caffe frame in the present embodiment.Realize the prototype network
Step includes:
Step 1: creation header file, addition layer header file is placed under include/caffe/layers/.Inherit in
vision_layers.hpp.The parameter of this layer is added in caffe.proto file.
Step 2: creating corresponding source file, it is placed under src/caffe/layers/, realizes LayerSetUp method,
For reading the parameter of layer, weight is initialized etc., and this method is called in layer::SetUp, is used for layer
Initialization.It realizes Reshape method, realize Forward_cpu and Backward_cpu method.
Step 3: creating .cu file at src/caffe/layers/ due to needing to accelerate using GPU.It realizes
Forward_gpu and Backward_gpu method.
Step 4: recompilating caffe code.
Wherein realizing that the convolutional layer needs the parameter designed includes the number of filter, the size of filter, parameter
Initial method, biasing initial method, whether open bias term, input it is every on one side plus how many a pixels and filter
Step-length.
As shown in Fig. 2, the step of training pattern, includes:
Step 1. pyrotechnics video acquisition；
Step 2. image preprocessing, the pyrotechnics video that specifically including will acquire are converted into the image sequence of one group of 10 frame, will
Pyrotechnics mark in image sequence is randomly assigned with the ratio of 1:3 with label at the image set for having label using classifier
Image set into test set and training set, the image in test set and training set is converted into LMDB format；
The pyrotechnics detection model that step 3. is reconstructed using training data training, until model is restrained；
Step 4. verifies pyrotechnics detection model using test data set.
When wherein training pyrotechnics detection model, as shown in Figure 1, the second full articulamentum (fc7) is by the in the prototype network
One full articulamentum (fc6) obtains after the convolution nuclear convolution of 1024 1*1, and the first full articulamentum (fc6) is by with first in VGG21
The complete connected pond layer (pool5) of articulamentum (fc6) obtains after the convolution nuclear convolution of 1024 3*3.In the prototype network
The 10th layer of convolutional layer (Conv4_3), the 15th layer of convolutional layer (Conv8_2), the 17th layer of convolutional layer (Conv9_2), the 19th layer volume
Lamination (Conv10_2), the 2nd layer of full articulamentum (fc7) and the 6th pond layer (pool11) using two 3*3 convolution kernels arranged side by side into
Row convolution, and then 6 groups of characteristic values are obtained, the confidence level of the prototype network final output is determined by this 6 groups of characteristic values.Wherein two
A convolution kernel arranged side by side is respectively to be used to detect the Localization of positioning target and for judging whether it is target
Confidence。
A kind of pyrotechnics identifying system has also been devised in the present invention, which includes video acquisition unit, image enhancement processing list
Member, pyrotechnics recognition unit, data storage unit, high-definition camera and warning device.
As shown in figure 3, the course of work of the system includes:
Step 1. video acquisition unit is for obtaining high-speed rail monitor video；
The insufficient picture of step 2. image enhancement processing cell processing illumination；
Step 3. carries out pyrotechnics detection using above-mentioned trained pyrotechnics detection model, when the confidence level of pyrotechnics detection is greater than
When 0.5, that is, it is determined as pyrotechnics；
Historical information when fire behavior occurs for the storage of step 4. data storage cell；
The alarm of step 5. warning device.
The foregoing is merely the preferred embodiments of the invention, are not intended to limit the invention creation, all at this
Within the spirit and principle of innovation and creation, any modification, equivalent replacement, improvement and so on should be included in the invention
Protection scope within.
Claims (7)
1. a kind of pyrotechnics recognizer based on SSD, which is characterized in that detection model used in pyrotechnics recognizer is by again
The prototype network training of design obtains, and the prototype network is reconstructed VGG16 network, and the prototype network is on the basis of VGG16
On reduce a full articulamentum, remain two full articulamentums, increase 6 convolutional layers and 1 pond layer.
2. a kind of pyrotechnics recognizer based on SSD as described in claim 1, which is characterized in that the step of training detection model
Include:
Step 1. pyrotechnics video acquisition；
Step 2. image preprocessing, the pyrotechnics video that specifically including will acquire is converted into the image sequence of one group of 10 frame, by image
Pyrotechnics mark in sequence is randomly assigned the figure with label using classifier at the image set for having label with the ratio of 1:3
Image in test set and training set is converted to LMDB format into test set and training set by image set；
The pyrotechnics detection model that step 3. is reconstructed using training data training, until model is restrained；
Step 4. verifies pyrotechnics detection model using test data set.
3. a kind of pyrotechnics recognizer based on SSD as claimed in claim 2, which is characterized in that when training pyrotechnics detection model,
The 10th layer of convolutional layer, the 15th layer of convolutional layer, the 17th layer of convolutional layer, the 19th layer of convolutional layer in the prototype network, the 2nd layer connect entirely
It connects layer and the 6th pond layer and carries out convolution using two 3*3 convolution kernels, obtain 6 groups of characteristic values.
4. a kind of pyrotechnics recognizer based on SSD as claimed in claim 3, which is characterized in that in described two 3*3 convolution kernels
One whether there is for detecting target, another is for judging whether target is pyrotechnics.
5. a kind of pyrotechnics recognizer based on SSD as claimed in claim 4, which is characterized in that when training pyrotechnics detection model,
The second full articulamentum is obtained after the convolution nuclear convolution of 1024 1*1 by the first full articulamentum in the prototype network, the mould
The first full articulamentum is obtained after the convolution nuclear convolution of 1024 3*3 by the pond layer being connected with the first full articulamentum in type network
It arrives.
6. a kind of pyrotechnics identifying system, which is characterized in that the pyrotechnics identifying system includes video acquisition unit, at image enhancement
Unit is managed, pyrotechnics recognition unit, data storage unit, wherein video acquisition unit increases for obtaining high-speed rail monitor video, image
For strong processing unit for handling the insufficient picture of illumination, the detection model training network by reconstruct is utilized in pyrotechnics recognition unit
Trained pyrotechnics detection model is determined as pyrotechnics, data storage cell is used for when the confidence level of pyrotechnics detection is greater than 0.5
Historical information when fire behavior occurs for storage.
7. a kind of pyrotechnics identifying system as claimed in claim 6, which is characterized in that the pyrotechnics identifying system further includes that high definition is taken the photograph
As head and warning device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810885395.6A CN109165575B (en) | 2018-08-06 | 2018-08-06 | Pyrotechnic recognition algorithm based on SSD frame |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810885395.6A CN109165575B (en) | 2018-08-06 | 2018-08-06 | Pyrotechnic recognition algorithm based on SSD frame |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109165575A true CN109165575A (en) | 2019-01-08 |
| CN109165575B CN109165575B (en) | 2024-02-20 |
Family
ID=64895071
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810885395.6A Active CN109165575B (en) | 2018-08-06 | 2018-08-06 | Pyrotechnic recognition algorithm based on SSD frame |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109165575B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110379136A (en) * | 2019-07-26 | 2019-10-25 | 北京讯腾智慧科技股份有限公司 | Supervisory control for gas leak device and method for combustion gas station |
| CN110390673A (en) * | 2019-07-22 | 2019-10-29 | 福州大学 | Cigarette automatic testing method based on deep learning under a kind of monitoring scene |
| CN111091551A (en) * | 2019-12-12 | 2020-05-01 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting loss fault of brake beam strut opening pin of railway wagon |
| CN111414969A (en) * | 2020-03-26 | 2020-07-14 | 西安交通大学 | Smoke detection method in foggy environment |
| CN112052744A (en) * | 2020-08-12 | 2020-12-08 | 成都佳华物链云科技有限公司 | Environment detection model training method, environment detection method and device |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101751744A (en) * | 2008-12-10 | 2010-06-23 | 中国科学院自动化研究所 | Detection and early warning method of smoke |
| US20110051993A1 (en) * | 2008-05-08 | 2011-03-03 | UTF Fire & Security | System and method for ensuring the performance of a video-based fire detection system |
| US20110058706A1 (en) * | 2008-05-08 | 2011-03-10 | Utc Fire & Secunity | System and method for video detection of smoke and flame |
| CN103956016A (en) * | 2014-04-25 | 2014-07-30 | 蚌埠依爱消防电子有限责任公司 | Smoke and flame combined type automatic detection method based on video image |
| CN106250845A (en) * | 2016-07-28 | 2016-12-21 | 北京智芯原动科技有限公司 | Flame detecting method based on convolutional neural networks and device |
| CN106845410A (en) * | 2017-01-22 | 2017-06-13 | 西安科技大学 | A kind of flame identification method based on deep learning model |
| CN107609470A (en) * | 2017-07-31 | 2018-01-19 | 成都信息工程大学 | The method of outdoor fire disaster early-stage smog video detection |
| CN107704820A (en) * | 2017-09-28 | 2018-02-16 | 深圳市鑫汇达机械设计有限公司 | A kind of effective coal-mine fire detecting system |
| CN107749067A (en) * | 2017-09-13 | 2018-03-02 | 华侨大学 | Fire hazard smoke detecting method based on kinetic characteristic and convolutional neural networks |
| CN107944359A (en) * | 2017-11-14 | 2018-04-20 | 中电数通科技有限公司 | Flame detecting method based on video |
| CN108052865A (en) * | 2017-07-06 | 2018-05-18 | 同济大学 | A kind of flame detecting method based on convolutional neural networks and support vector machines |
| CN108256496A (en) * | 2018-02-01 | 2018-07-06 | 江南大学 | A kind of stockyard smog detection method based on video |
| CN108257347A (en) * | 2018-01-10 | 2018-07-06 | 安徽大学 | A kind of flame image sequence sorting technique and device using convolutional neural networks |
-
2018
- 2018-08-06 CN CN201810885395.6A patent/CN109165575B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110051993A1 (en) * | 2008-05-08 | 2011-03-03 | UTF Fire & Security | System and method for ensuring the performance of a video-based fire detection system |
| US20110058706A1 (en) * | 2008-05-08 | 2011-03-10 | Utc Fire & Secunity | System and method for video detection of smoke and flame |
| CN101751744A (en) * | 2008-12-10 | 2010-06-23 | 中国科学院自动化研究所 | Detection and early warning method of smoke |
| CN103956016A (en) * | 2014-04-25 | 2014-07-30 | 蚌埠依爱消防电子有限责任公司 | Smoke and flame combined type automatic detection method based on video image |
| CN106250845A (en) * | 2016-07-28 | 2016-12-21 | 北京智芯原动科技有限公司 | Flame detecting method based on convolutional neural networks and device |
| CN106845410A (en) * | 2017-01-22 | 2017-06-13 | 西安科技大学 | A kind of flame identification method based on deep learning model |
| CN108052865A (en) * | 2017-07-06 | 2018-05-18 | 同济大学 | A kind of flame detecting method based on convolutional neural networks and support vector machines |
| CN107609470A (en) * | 2017-07-31 | 2018-01-19 | 成都信息工程大学 | The method of outdoor fire disaster early-stage smog video detection |
| CN107749067A (en) * | 2017-09-13 | 2018-03-02 | 华侨大学 | Fire hazard smoke detecting method based on kinetic characteristic and convolutional neural networks |
| CN107704820A (en) * | 2017-09-28 | 2018-02-16 | 深圳市鑫汇达机械设计有限公司 | A kind of effective coal-mine fire detecting system |
| CN107944359A (en) * | 2017-11-14 | 2018-04-20 | 中电数通科技有限公司 | Flame detecting method based on video |
| CN108257347A (en) * | 2018-01-10 | 2018-07-06 | 安徽大学 | A kind of flame image sequence sorting technique and device using convolutional neural networks |
| CN108256496A (en) * | 2018-02-01 | 2018-07-06 | 江南大学 | A kind of stockyard smog detection method based on video |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110390673A (en) * | 2019-07-22 | 2019-10-29 | 福州大学 | Cigarette automatic testing method based on deep learning under a kind of monitoring scene |
| CN110379136A (en) * | 2019-07-26 | 2019-10-25 | 北京讯腾智慧科技股份有限公司 | Supervisory control for gas leak device and method for combustion gas station |
| CN111091551A (en) * | 2019-12-12 | 2020-05-01 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting loss fault of brake beam strut opening pin of railway wagon |
| CN111414969A (en) * | 2020-03-26 | 2020-07-14 | 西安交通大学 | Smoke detection method in foggy environment |
| CN112052744A (en) * | 2020-08-12 | 2020-12-08 | 成都佳华物链云科技有限公司 | Environment detection model training method, environment detection method and device |
| CN112052744B (en) * | 2020-08-12 | 2024-02-09 | 成都佳华物链云科技有限公司 | Environment detection model training method, environment detection method and environment detection device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109165575B (en) | 2024-02-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109165575A (en) | A kind of pyrotechnics recognizer based on SSD frame | |
| CN109558832B (en) | Human body posture detection method, device, equipment and storage medium | |
| CN109344821A (en) | Small target detecting method based on Fusion Features and deep learning | |
| CN112861635B (en) | Fire disaster and smoke real-time detection method based on deep learning | |
| CN111079518B (en) | Ground-falling abnormal behavior identification method based on law enforcement and case handling area scene | |
| CN113850242A (en) | Storage abnormal target detection method and system based on deep learning algorithm | |
| CN109784278A (en) | The small and weak moving ship real-time detection method in sea based on deep learning | |
| CN107563299B (en) | Pedestrian detection method using RecNN to fuse context information | |
| CN112132205B (en) | Remote sensing image classification method based on convolutional neural network | |
| CN114882430B (en) | Transformer-based lightweight early fire detection method | |
| CN114519819B (en) | Remote sensing image target detection method based on global context awareness | |
| CN109271906A (en) | A kind of smog detection method and its device based on depth convolutional neural networks | |
| CN109376580A (en) | A kind of electric tower component identification method based on deep learning | |
| CN116824335A (en) | YOLOv5 improved algorithm-based fire disaster early warning method and system | |
| KR102540208B1 (en) | Fire detection method using deep learning | |
| Xu et al. | Concrete crack segmentation based on convolution–deconvolution feature fusion with holistically nested networks | |
| Sun et al. | Image recognition technology in texture identification of marine sediment sonar image | |
| CN108830172A (en) | Aircraft remote sensing images detection method based on depth residual error network and SV coding | |
| Yang et al. | Efficient Detection of Forest Fire Smoke in UAV Aerial Imagery Based on an Improved Yolov5 Model and Transfer Learning | |
| Wang et al. | SE-COTR: A novel fruit segmentation model for green apples application in complex orchard | |
| Sun et al. | UAV image detection algorithm based on improved YOLOv5 | |
| Zheng et al. | A lightweight algorithm capable of accurately identifying forest fires from UAV remote sensing imagery | |
| Tu et al. | An improved YOLOv5 for object detection in visible and thermal infrared images based on contrastive learning | |
| Yasser et al. | Egyart_classify: an approach to classify outpainted Egyptian monuments images using GAN and ResNet | |
| Li et al. | DAR‐Net: Dense Attentional Residual Network for Vehicle Detection in Aerial Images |
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 |