CN112712008A - Water environment early warning judgment method based on 3D convolutional neural network - Google Patents
Water environment early warning judgment method based on 3D convolutional neural network Download PDFInfo
- Publication number
- CN112712008A CN112712008A CN202011572402.0A CN202011572402A CN112712008A CN 112712008 A CN112712008 A CN 112712008A CN 202011572402 A CN202011572402 A CN 202011572402A CN 112712008 A CN112712008 A CN 112712008A
- Authority
- CN
- China
- Prior art keywords
- early warning
- neural network
- convolutional neural
- water quality
- video
- 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
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/18—Water
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Chemical & Material Sciences (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a water environment early warning judgment method based on a 3D convolutional neural network, which comprises the steps of extracting image frames from a shooting video and acquiring a swimming sequence corresponding to a model organism; preprocessing a mode biological swimming sequence, and dividing a training set and a test set; inputting the preprocessed mode biological swimming sequence into a 3D convolutional neural network for training to obtain a trained model capable of distinguishing different characteristics of mode biological swimming videos in normal water quality or abnormal water quality; inputting a test set monitoring video, and outputting the swimming characteristics of the model creatures after model identification; capturing biological characters of model organisms expressed in different water qualities to form a water environment early warning system based on a 3D convolutional neural network; and arranging the video streams acquired in real time according to a time sequence, inputting the video streams into the 3D convolutional neural network, and outputting a test result. The invention can accurately identify the abnormal water quality in real time and carry out early warning in time, thereby realizing the identification and prediction of the water environment.
Description
Technical Field
The invention relates to a water environment early warning and judging method based on a 3D convolutional neural network.
Background
The water resource is a necessary resource for human survival and development, the problem of water pollution in rivers, lakes and reservoirs is increasingly aggravated along with the development of modern industrialization in China, in order to ensure water safety and monitor water quality, a main means for controlling water pollution is provided, the dynamic change rule of the water environment is mastered by effectively monitoring and early warning the water environment, and the water pollution event can be responded in time so as to better improve the quality of life of residents.
The 3D convolution is used in a deep neural network system structure and is a very efficient method for learning video features, the 3D convolution neural network is developed from a 2D convolution neural network, the 2D convolution neural network only processes feature information on an image space domain, and for a video sequence, the 3D convolution neural network can simultaneously process information of a time domain by adding time dimensions on a convolution layer and a pooling layer, so that space-time information in a video is obtained.
The model organisms, such as daphnia magna, zebra fish and the like, are very sensitive to eutrophication and toxic substance water body dissolved oxygen in water quality, can indicate the water body eutrophication degree and the existence and range of toxic substances, can change differently under different dissolved oxygen concentrations, and are subjected to 3D convolution behavior analysis, so that the pollution degree of the water body is indicated, early warning alarm is timely made, and therefore, the state of the model organisms can be monitored in real time to achieve the monitoring of water quality safety.
Most of the traditional monitoring methods are to utilize physicochemical equipment or establish physicochemical indexes of water quality detected by a monitoring center to judge whether the water quality is abnormal, and the two methods cannot simultaneously meet the requirements of low input cost, simple construction and real-time online early warning.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an efficient and real-time water environment early warning judgment method, which combines a 3D convolutional neural network and water quality monitoring equipment to establish a water environment early warning system, performs characteristic extraction and analysis on a shot mode biological video, establishes a deep learning network to perform learning modeling, judges whether the water quality is abnormal or not by utilizing the trained 3D convolutional neural network, realizes the online monitoring of water quality safety, and realizes the timely prediction and early warning.
In order to achieve the above purpose, the invention is realized by the following technical scheme:
a water environment early warning judgment method based on a 3D convolutional neural network comprises the following steps:
s1: shooting a swimming video of a mode organism in normal water quality and a swimming video of the mode organism in abnormal water quality, extracting image frames from the videos, and acquiring a swimming sequence corresponding to the mode organism;
s2: preprocessing a mode biological swimming sequence, and dividing a training set and a test set;
s3: inputting the preprocessed mode biological swimming sequence into a 3D convolution neural network for training, and extracting time and space characteristics of the mode biological swimming sequence through a 3D convolution operation kernel to obtain a trained model capable of distinguishing different characteristics of mode biological swimming videos in normal water quality or abnormal water quality;
s4: inputting a test set monitoring video, and outputting the swimming characteristics of the model organisms after model identification so as to judge whether the water quality is abnormal;
s5: capturing biological characters of model organisms expressed in different water qualities to form a water environment early warning system based on a 3D convolutional neural network;
s6: when the early warning system is applied to water environment of water quality condition for judgment, the classification and early warning of water quality can be realized only by arranging and inputting real-time acquired video streams into the 3D convolutional neural network according to time sequence;
s7: and outputting a test result, and timely early warning the identified abnormal water quality.
Further, the image frame extracted from the video is a key frame of the video.
Further, in S2, the model biological swimming sequence preprocessing packet image is randomly cropped, scaled, normalized, and background differentiated to meet the requirements of uniform definition, size, and brightness.
Further, the cross entropy loss function calculation of the 3D convolutional neural network training includes:
wherein Loss represents cross entropy Loss function, N represents number of samples, yiThe actual value is represented by the value of,indicating the predicted value.
Further, in S5, the water environment early warning system includes the following four modules:
a) the video shooting module: shooting a water environment video in real time;
b) an image preprocessing module: for preprocessing the collected images;
c) a 3D convolutional neural network module: performing 3D convolution and maximum pooling operation on time and space dimensions, capturing continuous change information obtained from continuous time periods, and finally combining all obtained information to obtain final feature description;
d) the video network transmission and prediction module in practical application comprises: and in practical application, the real-time monitoring video is transmitted to the terminal, the water quality condition is judged according to characteristic comparison, and a prediction result is made on the water environment.
Furthermore, the practical application of the early warning system is that the early warning system is applied to monitoring the water quality condition of the lake and reservoir drainage basin, the real-time condition of the water quality of the lake and reservoir drainage basin is analyzed and judged, and when the system monitors that the mode biological swimming track is combined with the abnormal water quality characteristic, an alarm is given out to remind a worker to take a treatment measure.
The invention has the beneficial effects that:
1. the biological characters of the model organisms expressed in different water qualities are captured and distinguished, so that whether the water quality is abnormal or not can be identified, the method is used for abnormity early warning of water quality monitoring, and the method has great significance for establishing and perfecting a water resource environment monitoring system and enhancing monitoring capacity construction for preventing and treating water pollution.
2. By adopting the mode biological early warning, the water quality change characteristics of the drainage basin can be reflected, the timeliness of the early warning can be guaranteed, and the early warning performance-price ratio is high.
3. The water environment early warning system is built by using the 3D convolutional neural network, features are automatically learned from the water environment early warning system, a deep learning model with multiple hidden layers is built, accurate and effective features are automatically obtained by a computer, and finally the precision of the water environment early warning system is improved.
Drawings
Fig. 1 is a schematic overall flow chart of a water environment early warning and judging method based on a 3D convolutional neural network.
Fig. 2 is a schematic structural diagram of a 3D convolutional neural network in the water environment early warning and judging method based on the 3D convolutional neural network of the present invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings, but the invention is not limited thereto.
As shown in fig. 1 and 2, a water environment early warning and judging method based on a 3D convolutional neural network mainly includes the following steps:
s1: shooting a swimming video of a mode organism in normal water quality and a swimming video of the mode organism in abnormal water quality, extracting image frames from the videos, and acquiring a swimming sequence corresponding to the mode organism;
s2: preprocessing a mode biological swimming sequence, and dividing a training set and a test set;
s3: inputting the preprocessed mode biological swimming sequence into a 3D convolution neural network for training, and extracting time and space characteristics of the mode biological swimming sequence through a 3D convolution operation kernel to obtain a trained model capable of distinguishing different characteristics of mode biological swimming videos in normal water quality or abnormal water quality;
s4: inputting a test set monitoring video, and outputting the swimming characteristics of the model organisms after model identification so as to judge whether the water quality is abnormal;
s5: capturing biological characters of model organisms expressed in different water qualities to form a water environment early warning system based on a 3D convolutional neural network;
s6: when the early warning system is applied to the water environment with unknown water quality condition for judgment, the classification and early warning of the water quality can be realized only by arranging the video streams acquired in real time according to the time sequence and inputting the video streams into the 3D convolutional neural network;
s7: and outputting a test result, and timely early warning the identified abnormal water quality.
Example 1
As shown in fig. 1 and 2, this embodiment will further describe the present invention with reference to specific data, where the data is optimized data, the data set of this embodiment has 1000 video segments, the length of each video segment is about 20s, and the total duration of the video segments is 5.6 h.
Step 1: shooting the swimming video of the model organism in normal water quality and the swimming video of the model organism in abnormal water quality, wherein the abnormal water quality has heavy metal chromium ions (Cr)6+) Excess of mercury ions (Hg)2+) And (3) exceeding the standard, and the content of oscillatoria algae toxins and organic solvents exceeding the standard, and the like, extracting image frames from 1000 sections of the shot moving video of the model organisms, and acquiring the moving sequence corresponding to the model organisms.
Step 2: the extracted 1000 video sequences are segmented into frames according to the time sequence, and the re-normalization processing is carried out according to the length and the width of the input format, for example, each frame of picture is required to be in a 'jpg' format, the length and the width of the picture are 128 multiplied by 171, each frame of picture is cut out, and a training set and a testing set are divided.
And step 3: the segmented pattern biological swimming video frame picture sequence is input into a 3D convolutional neural network for training, a 3D convolutional neural network model used in the embodiment is shown in FIG. 2, and comprises 6 convolutional layers, 6 pooling layers and 2 full-connection layers, the input of the network model is a processed pattern biological swimming sequence, and the output is a water environment early warning system based on the 3D convolutional neural network.
Computing the true value y using a cross-entropy loss function trained by a 3D convolutional neural networkiAnd the predicted valueThe specific form of the deviation is as follows:
wherein Loss represents cross entropy Loss function, N represents number of samples, yiThe actual value is represented by the value of,indicating the predicted value.
And 4, step 4: and testing and verifying different characteristic models of the model organisms obtained by training in the steps in different water qualities, inputting a test monitoring video and outputting an early warning type.
And 5: when the early warning system is applied to the water environment with unknown water quality condition for judgment, the video streams acquired in real time are input into the 3D convolutional neural network in a time sequence, when the system monitors that the input information accords with the abnormal water quality change characteristics, an alarm is given out, and workers are reminded to take treatment measures in time, so that the guarantee is provided for water quality health and safe water use.
The above-mentioned embodiments are merely preferred embodiments of the present invention, and the present invention is not limited thereto, and various changes and modifications can be made by workers without departing from the spirit and principle of the present invention, and any modifications, equivalents, improvements, etc. made by workers according to the present invention shall fall within the protection scope of the present invention.
Claims (6)
1. A water environment early warning judgment method based on a 3D convolutional neural network is characterized by comprising the following steps:
s1: shooting a swimming video of a mode organism in normal water quality and a swimming video of the mode organism in abnormal water quality, extracting image frames from the videos, and acquiring a swimming sequence corresponding to the mode organism;
s2: preprocessing a mode biological swimming sequence, and dividing a training set and a test set;
s3: inputting the preprocessed mode biological swimming sequence into a 3D convolution neural network for training, and extracting time and space characteristics of the mode biological swimming sequence through a 3D convolution operation kernel to obtain a trained model capable of distinguishing different characteristics of mode biological swimming videos in normal water quality or abnormal water quality;
s4: inputting a test set monitoring video, and outputting the swimming characteristics of the model organisms after model identification so as to judge whether the water quality is abnormal;
s5: capturing biological characters of model organisms expressed in different water qualities to form a water environment early warning system based on a 3D convolutional neural network;
s6: when the early warning system is applied to water environment of water quality condition for judgment, the classification and early warning of water quality can be realized only by arranging and inputting real-time acquired video streams into the 3D convolutional neural network according to time sequence;
s7: and outputting a test result, and timely early warning the identified abnormal water quality.
2. The water environment early warning and judging method based on the 3D convolutional neural network as claimed in claim 1, which is characterized in that:
the image frames extracted from the video are key frames of the video.
3. The water environment early warning and judging method based on the 3D convolutional neural network as claimed in claim 1, which is characterized in that:
in S2, the mode biological swimming sequence preprocessing package image is cut randomly, scaled, normalized and background differentiated to meet the requirement of uniform definition, size and brightness.
4. The water environment early warning and judging method based on the 3D convolutional neural network as claimed in claim 1, which is characterized in that: the cross entropy loss function computation of the 3D convolutional neural network training comprises:
5. The water environment early warning and judging method based on the 3D convolutional neural network as claimed in claim 1, which is characterized in that: in S5, the water environment early warning system includes the following four modules:
a) the video shooting module: shooting a water environment video in real time;
b) an image preprocessing module: for preprocessing the collected images;
c) a 3D convolutional neural network module: performing 3D convolution and maximum pooling operation on time and space dimensions, capturing continuous change information obtained from continuous time periods, and finally combining all obtained information to obtain final feature description;
d) the video network transmission and prediction module in practical application comprises: and in practical application, the real-time monitoring video is transmitted to the terminal, the water quality condition is judged according to characteristic comparison, and a prediction result is made on the water environment.
6. The water environment early warning and judging method based on the 3D convolutional neural network as claimed in claim 1, which is characterized in that: in S6, the early warning system is applied to monitoring water quality of lake and reservoir areas, analyzing and judging the real-time water quality of the lake and reservoir areas, and when the system monitors that the mode biological swimming trajectory is combined with abnormal water quality characteristics, giving an alarm to remind workers to take treatment measures.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011572402.0A CN112712008A (en) | 2020-12-28 | 2020-12-28 | Water environment early warning judgment method based on 3D convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011572402.0A CN112712008A (en) | 2020-12-28 | 2020-12-28 | Water environment early warning judgment method based on 3D convolutional neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112712008A true CN112712008A (en) | 2021-04-27 |
Family
ID=75545587
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011572402.0A Pending CN112712008A (en) | 2020-12-28 | 2020-12-28 | Water environment early warning judgment method based on 3D convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112712008A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111062316A (en) * | 2019-12-16 | 2020-04-24 | 成都之维安科技股份有限公司 | Pollution source wastewater discharge real-time video analysis system based on deep learning technology |
CN115457485A (en) * | 2022-11-11 | 2022-12-09 | 成都见海科技有限公司 | Drainage monitoring method and system based on 3D convolution and storage medium |
-
2020
- 2020-12-28 CN CN202011572402.0A patent/CN112712008A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111062316A (en) * | 2019-12-16 | 2020-04-24 | 成都之维安科技股份有限公司 | Pollution source wastewater discharge real-time video analysis system based on deep learning technology |
CN115457485A (en) * | 2022-11-11 | 2022-12-09 | 成都见海科技有限公司 | Drainage monitoring method and system based on 3D convolution and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gu et al. | Deep dual-channel neural network for image-based smoke detection | |
CN107123131B (en) | Moving target detection method based on deep learning | |
CN111738044B (en) | Campus violence assessment method based on deep learning behavior recognition | |
CN112712008A (en) | Water environment early warning judgment method based on 3D convolutional neural network | |
CN108305243A (en) | A kind of magnetic tile surface defect detection method based on deep learning | |
CN107909564A (en) | A kind of full convolutional network image crack detection method based on deep learning | |
CN112465748A (en) | Neural network based crack identification method, device, equipment and storage medium | |
CN111896540B (en) | Water quality on-line monitoring system based on block chain | |
Yu et al. | Robust median filtering forensics by CNN-based multiple residuals learning | |
Shi et al. | Research on the membrane fouling diagnosis of MBR membrane module based on ECA-CNN | |
CN115526852A (en) | Molten pool and splash monitoring method in selective laser melting process based on target detection and application | |
CN116311214A (en) | License plate recognition method and device | |
CN114241423A (en) | Intelligent detection method and system for river floaters | |
Gupta et al. | Aquasight: Automatic water impurity detection utilizing convolutional neural networks | |
CN112686801A (en) | Water quality monitoring method based on aerial image and series echo state network | |
CN115296933B (en) | Industrial production data risk level assessment method and system | |
CN116704444A (en) | Video abnormal event detection method based on cascade attention U-Net | |
CN112818948B (en) | Behavior identification method based on visual attention under embedded system | |
CN115830309A (en) | SGAN road crack image segmentation method based on Deeplabv3+ | |
CN114005064A (en) | Biological water body pollution early warning method and device based on machine vision technology | |
Cao et al. | HCHODetector: Formaldehyde concentration detection based on deep learning | |
CN111443053A (en) | Water quality on-line monitoring and early warning system based on biological behavior and multispectral | |
Ye et al. | Prediction of the Cyanobacteria Coverage in Time-series Images based on Convolutional Neural Network | |
CN117787570B (en) | Comprehensive evaluation method and system for heavy metal pollution of water environment | |
CN117764993B (en) | Water quality on-line monitoring system and method based on image analysis |
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 |