CN112580611A - Air pollution assessment method based on IGAN-CNN model - Google Patents
Air pollution assessment method based on IGAN-CNN model Download PDFInfo
- Publication number
- CN112580611A CN112580611A CN202110194564.3A CN202110194564A CN112580611A CN 112580611 A CN112580611 A CN 112580611A CN 202110194564 A CN202110194564 A CN 202110194564A CN 112580611 A CN112580611 A CN 112580611A
- Authority
- CN
- China
- Prior art keywords
- image
- model
- air pollution
- cnn
- igan
- 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
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
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Abstract
An air pollution assessment method based on an IGAN-CNN model. Step 1, acquiring a satellite image: step 2, image enhancement processing: and 3, data sample expansion: carrying out sample expansion on the satellite image in the step 2 by utilizing the generated countermeasure network; step 4, off-line training of the model: and 5, online application of the model: and (4) applying the IGAN-CNN model obtained by training in the steps 1-4 on line to realize the evaluation of the air pollution degree of the image shot by the satellite in real time. The method provided by the invention has good evaluation precision and generalization performance, and has good practical application value.
Description
Technical Field
The invention relates to the field of air pollution degree evaluation, in particular to an air pollution evaluation method based on an IGAN-CNN model.
Background
China is the world with the greatest development, the economic and social development is highly effective in recent years, the urbanization process is steadily promoted, the industrialization degree is continuously improved, and the living standard of people is comprehensively improved and promoted. However, the air pollution problem caused by historical accumulation is increasingly prominent and becomes a key factor for restricting the sustainable development of China. In order to get rid of the air pollution predicament, the Chinese government actively controls the air pollution by using financial and administrative means, and the environmental protection tax Law is formally implemented in 2018; the domestic scholars also respectively develop researches on the urban air pollution problem from the aspects of spatial distribution, emission reduction measures, influence factors, development modes, monitoring technologies and the like, and provide corresponding improvement and solution measures.
How to quickly and accurately evaluate the current air pollution degree is a very meaningful research topic, and deep learning is taken as a basic technology of the current most popular artificial intelligence field, and the capability of analyzing and processing big data is particularly excellent. Deep learning can screen and extract features of a large amount of data information such as characters, images or sounds through a deep and multi-level artificial neural network structure, and learn the feature representation of a specific object, so that high-level processing of data is completed after a large amount of data information is accurately understood. Therefore, compared with the traditional satellite image classification and identification based on image processing, the satellite image classification and identification method based on the deep learning technology can be used for carrying out classification and identification on the air pollution degree through the satellite image in a targeted manner by aiming at the current massive satellite images.
The Chinese invention patent related to classifying and identifying the air pollution degree in China has an air pollution estimation method based on satellite images, CN201910238150.9, No. CN109978862A, No. 20190907, wherein the infrared image is subjected to threshold value filtration, then the image subjected to threshold value filtration is subjected to key point detection, the overlapping rate is calculated, and the obtained detection key points are deduplicated to obtain scattered detection key points and detection ranges; then, threshold filtering is performed on the images of the dispersed detection key points and the detection range, the detection key points with pixel values lower than the threshold are removed, and the remaining detection key points and the detection range are restored in the original image. The invention relates to a method and a system for detecting an air pollution index according to a mobile terminal, in particular to a method and a system for detecting an air pollution index according to the mobile terminal, which are disclosed in China, wherein the method and the system are CN201510096538.1, No. CN104614295A, No. 20150513, the concentration of current suspended particles in the air is detected by a suspended particle sensing device on the mobile terminal, the detected concentration of the current suspended particles is sent to a terminal processor for processing, and the concentration grade of the current suspended particles is obtained.
Disclosure of Invention
In order to solve the problems, the invention provides an air pollution assessment method based on an IGAN-CNN model on the basis of a neighborhood average algorithm and GAN and CNN network models. Considering that the image shot by the satellite has noise, the method adopts a neighborhood average algorithm to perform noise reduction processing on the image so as to highlight the output characteristics of the image; in order to improve the generalization performance of the model, the GAN model is adopted to expand the satellite images so as to increase the training sample size of the model. In addition, a CNN model with high expression capacity is adopted to carry out deep feature mining on satellite image data, a dropout technology is used in the CNN model to prevent a network model from being over-fitted, and finally accurate classification and identification of satellite images are achieved.
To achieve the purpose, the invention provides an air pollution assessment method based on an IGAN-CNN model, which comprises the following specific steps:
step 1, acquiring a satellite image: shooting thermal infrared detection images under different air pollution degrees by using a satellite, wherein a 10-TIRS shooting waveband is selected;
step 2, image enhancement processing: denoising the image in the step 1 by utilizing a neighborhood average algorithm to enhance the output purity of the image;
and 3, data sample expansion: carrying out sample expansion on the satellite image in the step 2 by utilizing a generated countermeasure network GAN;
step 4, off-line training of the model: making a corresponding label for the image obtained in the step 3, and inputting the label into a Convolutional Neural Network (CNN) for model training until the model converges;
and 5, online application of the model: and (4) applying the IGAN-CNN obtained by training in the steps 1-4 to the offline model on line to realize the evaluation of the air pollution degree of the image shot by the satellite in real time.
Further, the classification of the degree of air pollution in step 1 is specifically described as follows:
according to the forecast grade standard and the implementation scheme of the air pollution meteorological condition forecast service, the air pollution degree is divided into 6 grades, namely: good first-level representation, good second-level representation, general third-level representation, poor fourth-level representation, poor fifth-level representation, and extremely poor sixth-level representation.
Further, the specific description of the enhancement processing of the image by using the neighborhood average algorithm in the step 2 is as follows:
suppose that the images taken by the satellites are represented asf(a,b) The length and width of the image beingLAndDin eight neighborhood noise reduction, pixel pointsf(i,j) The eight neighborhoods of (a) may be expressed as:
image obtained after noise reduction of satellite image by neighborhood averaging methodp(a,b) Can be expressed as:
in the formula
Representing by pixel pointsf(a,b) A set of neighborhoods that are centered,nis the total number in the set, and 1 ≦a≤L-1，1≤b≤D-1。
Further, the specific steps of performing sample expansion on the satellite image by using GAN in the step 3 are as follows:
step 3.1, constructing a generator G and a recognizer D framework in the GAN, wherein G and D in the patent are both CNN network models;
step 3.2, fixing one model to train the other model, namely: D/G is trained by fixed G/D, G and D are continuously promoted in the training process respectively, and the objective function of the countermeasure processV(D,G) Can be expressed as:
in the formula (I), the compound is shown in the specification,
representing satellite imagesxThe probability distribution of (a) is determined,
representing generated extended sampleszThe probability distribution of (a) is determined,D(x) Representing the output of the satellite image after D,G(z) Representing extended sampleszOutputting after G;
step 3.3, updating parameters of D and G by using a random gradient descent SGD algorithm, wherein the updating criterion can be expressed as:
in the formula (I), the compound is shown in the specification,
and
respectively, D is a parameter of G,mrepresents the total amount of samples trained;
and 3.4, repeating the steps 3.2-3.3 until Nash balance is reached, and considering that the expanded sample probability distribution is consistent with the probability distribution of the original image.
Further, the specific step of training the off-line model for the IGAN-CNN in step 4 is as follows:
step 4.1, a CNN network model is built by using TensorFlow, wherein the network structure is as follows: the method comprises the following steps of inputting a layer, namely a convolution layer 1, a pooling layer 1, a dropout layer 1, a convolution layer 2, a pooling layer 2, a dropout layer 2, a convolution layer 3, a pooling layer 3, a dropout layer 3, a full-connection layer 1, a full-connection layer 2 and a softmax layer;
step 4.2, inputting the obtained image into a CNN for training, wherein a cross entropy loss function is adopted as a loss function, and the specific expression is as follows:
in the formula (I), the compound is shown in the specification,
in order to be the actual sample label,
a label for discriminating the Softmax layer;
step 4.3, updating the parameters in the CNN by using the SGD until the loss functionLThe set convergence threshold 1e-5 is reached.
The air pollution evaluation method based on the IGAN-CNN model has the beneficial effects that: the invention has the technical effects that:
1. the method utilizes the neighborhood average algorithm to perform noise reduction processing on the image shot by the satellite, thereby well highlighting the output characteristics of the image, and having great effect on improving the identification precision of the model;
2. according to the method, the training samples of the model are expanded by utilizing the GAN, so that the problems that the deep learning model is easy to over-fit and insufficient in generalization under the condition of small sample training are greatly improved;
3. the method adopts the CNN model with strong nonlinear expression capability to carry out deep feature mining on the satellite image data, and uses a dropout technology in the CNN model to prevent the network model from being over-fitted, thereby finally realizing accurate classification and identification on the satellite image.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a network structure diagram of the IGAN-CNN model used in the present invention.
Detailed Description
The invention is described in further detail below with reference to the following detailed description and accompanying drawings:
the invention provides an air pollution evaluation method based on an IGAN-CNN model, aiming at accurately and effectively evaluating the air pollution degree by using an image shot by a satellite. FIG. 1 is a flow chart of the present invention. The steps of the present invention will be described in detail with reference to the flow chart.
Step 1, acquiring a satellite image: shooting thermal infrared detection images under different air pollution degrees by using a Landsat-8 satellite, wherein a 10-TIRS shooting waveband is selected;
the classification of the air pollution degree in step 1 is specifically described as follows:
according to the air pollution weather condition forecast rating standard and the air pollution weather condition forecast service implementation scheme (2013-: first (good), second (good), third (normal), fourth (poor), fifth (poor) and sixth (poor).
Step 2, image enhancement processing: denoising the image in the step 1 by utilizing a neighborhood average algorithm to enhance the output purity of the image;
the specific description of the enhancement processing of the image by using the neighborhood average algorithm in the step 2 is as follows:
suppose that the images taken by the satellites are represented asf(a,b) The length and width of the image beingLAndDin eight neighborhood noise reduction, pixel pointsf(i,j) The eight neighborhoods of (a) may be expressed as:
image obtained after noise reduction of satellite image by neighborhood averaging methodp(a,b) Can be expressed as:
in the formula
Representing by pixel pointsf(a,b) A set of neighborhoods that are centered,nis the total number in the set, and 1 ≦a≤L-1，1≤b≤D-1。
And 3, data sample expansion: carrying out sample expansion on the satellite image in the step 2 by utilizing a generation countermeasure network (GAN);
the specific steps of utilizing GAN to carry out sample expansion on the satellite image in the step 3 are as follows:
step 3.1, constructing a generator G and a recognizer D framework in the GAN, wherein G and D in the patent are both CNN network models;
step 3.2, fixing one model to train the other model, namely: D/G is trained by fixed G/D, G and D are continuously promoted in the training process respectively, and the objective function of the countermeasure processV(D,G) Can be expressed as:
in the formula (I), the compound is shown in the specification,
representing satellite imagesxThe probability distribution of (a) is determined,
representing generated extended sampleszThe probability distribution of (a) is determined,D(x) Representing the output of the satellite image after D,G(z) Representing extended sampleszOutputting after G;
step 3.3, updating parameters of D and G by using a random gradient descent SGD algorithm, wherein the updating criterion can be expressed as:
in the formula (I), the compound is shown in the specification,
and
respectively, D is a parameter of G,mrepresents the total amount of samples trained;
and 3.4, repeating the steps 3.2-3.3 until Nash balance is reached, and considering that the expanded sample probability distribution is consistent with the probability distribution of the original image.
Step 4, off-line training of the model: making a corresponding label for the image obtained in the step 3, and inputting the label into a Convolutional Neural Network (CNN) for model training until the model converges;
the specific steps of training the IGAN-CNN to the offline model in the step 4 are as follows:
step 4.1, a CNN network model is built by using TensorFlow, wherein the network structure is as follows: the method comprises the following steps of inputting a layer, namely a convolution layer 1, a pooling layer 1, a dropout layer 1, a convolution layer 2, a pooling layer 2, a dropout layer 2, a convolution layer 3, a pooling layer 3, a dropout layer 3, a full-connection layer 1, a full-connection layer 2 and a softmax layer;
step 4.2, inputting the obtained image into a CNN for training, wherein a cross entropy loss function is adopted as a loss function, and the specific expression is as follows:
in the formula (I), the compound is shown in the specification,
in order to be the actual sample label,
a label for discriminating the Softmax layer;
step 4.3, updating the parameters in the CNN by using the SGD until the loss functionLThe set convergence threshold 1e-5 is reached. And 5, online application of the model: and (4) applying the IGAN-CNN model obtained by training in the steps 1-4 on line to realize the evaluation of the air pollution degree of the image shot by the satellite in real time.
Fig. 2 is a network structure diagram of the IGAN-CNN model used in the present invention. It is clear from the network structure diagram that the IGAN-CNN model consists of five major parts: the method comprises the following steps of shooting thermal infrared detection images under different air pollution degrees by using a Landsat-8 satellite; then, filtering and enhancing the image by utilizing a neighborhood average algorithm; and then, carrying out sample expansion on the acquired satellite images by using a GAN network, wherein a generator G and a recognizer D in the GAN are alternately trained, namely: fixing G/D training D/G; then, utilizing the shot image and the expanded image to perform offline training on the CNN model until the model converges; and finally, applying the offline trained model on line to realize the evaluation of the air pollution degree.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, but any modifications or equivalent variations made according to the technical spirit of the present invention are within the scope of the present invention as claimed.
Claims (5)
1. An air pollution assessment method based on an IGAN-CNN model comprises the following specific steps:
step 1, acquiring a satellite image: shooting thermal infrared detection images under different air pollution degrees by using a satellite, wherein a 10-TIRS shooting waveband is selected;
step 2, image enhancement processing: denoising the image in the step 1 by utilizing a neighborhood average algorithm to enhance the output purity of the image;
and 3, data sample expansion: carrying out sample expansion on the satellite image in the step 2 by utilizing a generated countermeasure network GAN;
step 4, off-line training of the model: making a corresponding label for the image obtained in the step 3, and inputting the label into a Convolutional Neural Network (CNN) for model training until the model converges;
and 5, online application of the model: and (4) applying the IGAN-CNN obtained by training in the steps 1-4 to the offline model on line to realize the evaluation of the air pollution degree of the image shot by the satellite in real time.
2. The IGAN-CNN model-based air pollution assessment method according to claim 1, wherein: the classification of the air pollution degree in step 1 is specifically described as follows:
according to the forecast grade standard and the implementation scheme of the air pollution meteorological condition forecast service, the air pollution degree is divided into 6 grades, namely: good first-level representation, good second-level representation, general third-level representation, poor fourth-level representation, poor fifth-level representation, and extremely poor sixth-level representation.
3. The IGAN-CNN model-based air pollution assessment method according to claim 1, wherein: the specific description of the enhancement processing of the image by using the neighborhood average algorithm in the step 2 is as follows:
suppose that the images taken by the satellites are represented asf(a,b) The length and width of the image beingLAndDin eight neighborhood noise reduction, pixel pointsf(i,j) The eight neighborhoods of (a) may be expressed as:
image obtained after noise reduction of satellite image by neighborhood averaging methodp(a,b) Can be expressed as:
in the formula
Representing by pixel pointsf(a,b) A set of neighborhoods that are centered,nis the total number in the set, and 1 ≦a≤L-1，1≤b≤D-1。
4. The IGAN-CNN model-based air pollution assessment method according to claim 1, wherein: the specific steps of utilizing GAN to carry out sample expansion on the satellite image in the step 3 are as follows:
step 3.1, constructing a generator G and a recognizer D framework in the GAN, wherein G and D in the patent are both CNN network models;
step 3.2, fixing one model to train the other model, namely: D/G is trained by fixed G/D, G and D are continuously promoted in the training process respectively, and the objective function of the countermeasure processV(D,G) Can be expressed as:
in the formula (I), the compound is shown in the specification,
representing satellite imagesxThe probability distribution of (a) is determined,
representing generated extended sampleszThe probability distribution of (a) is determined,D(x) Representing the output of the satellite image after D,G(z) Representing extended sampleszOutputting after G;
step 3.3, updating parameters of D and G by using a random gradient descent SGD algorithm, wherein the updating criterion can be expressed as:
in the formula (I), the compound is shown in the specification,
and
respectively, D is a parameter of G,mrepresents the total amount of samples trained;
and 3.4, repeating the steps 3.2-3.3 until Nash balance is reached, and considering that the expanded sample probability distribution is consistent with the probability distribution of the original image.
5. The IGAN-CNN model-based air pollution assessment method according to claim 1, wherein: the specific steps of training the IGAN-CNN to the offline model in the step 4 are as follows:
step 4.1, a CNN network model is built by using TensorFlow, wherein the network structure is as follows: the method comprises the following steps of inputting a layer, namely a convolution layer 1, a pooling layer 1, a dropout layer 1, a convolution layer 2, a pooling layer 2, a dropout layer 2, a convolution layer 3, a pooling layer 3, a dropout layer 3, a full-connection layer 1, a full-connection layer 2 and a softmax layer;
step 4.2, inputting the obtained image into a CNN for training, wherein a cross entropy loss function is adopted as a loss function, and the specific expression is as follows:
in the formula (I), the compound is shown in the specification,
in order to be the actual sample label,
a label for discriminating the Softmax layer;
step 4.3, updating the parameters in the CNN by using the SGD until the loss functionLThe set convergence threshold 1e-5 is reached.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110194564.3A CN112580611A (en) | 2021-02-21 | 2021-02-21 | Air pollution assessment method based on IGAN-CNN model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110194564.3A CN112580611A (en) | 2021-02-21 | 2021-02-21 | Air pollution assessment method based on IGAN-CNN model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112580611A true CN112580611A (en) | 2021-03-30 |
Family
ID=75113925
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110194564.3A Pending CN112580611A (en) | 2021-02-21 | 2021-02-21 | Air pollution assessment method based on IGAN-CNN model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112580611A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127702A (en) * | 2016-06-17 | 2016-11-16 | 兰州理工大学 | A kind of image mist elimination algorithm based on degree of depth study |
| CN107909084A (en) * | 2017-11-15 | 2018-04-13 | 电子科技大学 | A kind of haze concentration prediction method based on convolution linear regression network |
| CN108376397A (en) * | 2018-01-18 | 2018-08-07 | 中南大学 | A method of air quality is estimated based on single image |
| US20190236614A1 (en) * | 2018-01-29 | 2019-08-01 | Accenture Global Solutions Limited | Artificial intelligence counterfeit detection |
| CN111401261A (en) * | 2020-03-18 | 2020-07-10 | 金陵科技学院 | Robot gesture recognition method based on GAN-CNN framework |
-
2021
- 2021-02-21 CN CN202110194564.3A patent/CN112580611A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127702A (en) * | 2016-06-17 | 2016-11-16 | 兰州理工大学 | A kind of image mist elimination algorithm based on degree of depth study |
| CN107909084A (en) * | 2017-11-15 | 2018-04-13 | 电子科技大学 | A kind of haze concentration prediction method based on convolution linear regression network |
| CN108376397A (en) * | 2018-01-18 | 2018-08-07 | 中南大学 | A method of air quality is estimated based on single image |
| US20190236614A1 (en) * | 2018-01-29 | 2019-08-01 | Accenture Global Solutions Limited | Artificial intelligence counterfeit detection |
| CN111401261A (en) * | 2020-03-18 | 2020-07-10 | 金陵科技学院 | Robot gesture recognition method based on GAN-CNN framework |
Non-Patent Citations (4)
| Title |
|---|
| 许庆勇: "《基于深度学习理论的纹身图像识别与检测研究》", 31 December 2018, 华中科技大学出版社 * |
| 陆系群等: "《图像处理原理、技术与算法》", 31 August 2001, 浙江大学出版社 * |
| 韩殿元: "《简单邻域平均图像去噪算法的改进研究》", 《潍坊学院学报》 * |
| 高扬: "《人工智能与机器人先进技术丛书 智能摘要与深度学习》", 31 July 2019, 北京理工大学出版社 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106650767B (en) | Flood forecasting method based on cluster analysis and real-time correction | |
| CN108229550B (en) | Cloud picture classification method based on multi-granularity cascade forest network | |
| CN109389180A (en) | A power equipment image-recognizing method and inspection robot based on deep learning | |
| CN108074244B (en) | Safe city traffic flow statistical method integrating deep learning and background difference method | |
| CN103198493B (en) | A kind ofly to merge and the method for tracking target of on-line study based on multiple features self-adaptation | |
| CN111709397B (en) | Unmanned aerial vehicle variable-size target detection method based on multi-head self-attention mechanism | |
| CN103605972A (en) | Non-restricted environment face verification method based on block depth neural network | |
| CN109919921B (en) | Environmental impact degree modeling method based on generation countermeasure network | |
| CN114333070A (en) | Examinee abnormal behavior detection method based on deep learning | |
| CN114255403A (en) | Optical remote sensing image data processing method and system based on deep learning | |
| CN116699096B (en) | Water quality detection method and system based on deep learning | |
| CN114169445A (en) | Day-ahead photovoltaic power prediction method, device and system based on CAE and GAN hybrid network | |
| CN115131747A (en) | Knowledge distillation-based power transmission channel engineering vehicle target detection method and system | |
| Wu et al. | Single shot multibox detector for vehicles and pedestrians detection and classification | |
| CN111242028A (en) | Remote sensing image ground object segmentation method based on U-Net | |
| CN109034192B (en) | Track-vehicle body vibration state prediction method based on deep learning | |
| CN114882373A (en) | Multi-feature fusion sandstorm prediction method based on deep neural network | |
| Li et al. | Automation recognition of pavement surface distress based on support vector machine | |
| CN113987910A (en) | Method and device for identifying load of residents by coupling neural network and dynamic time planning | |
| CN113343123A (en) | Training method and detection method for generating confrontation multiple relation graph network | |
| CN113283467A (en) | Weak supervision picture classification method based on average loss and category-by-category selection | |
| CN112507881A (en) | sEMG signal classification method and system based on time convolution neural network | |
| CN115272776B (en) | Hyperspectral image classification method based on double-path convolution and double attention and storage medium | |
| CN112580611A (en) | Air pollution assessment method based on IGAN-CNN model | |
| CN103279769B (en) | A kind of characteristics of objects expression of doing more physical exercises being applicable to different scene |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210330 |