CN112132089A - Excavator behavior analysis method based on 3D convolution and optical flow - Google Patents
Excavator behavior analysis method based on 3D convolution and optical flow Download PDFInfo
- Publication number
- CN112132089A CN112132089A CN202011054904.4A CN202011054904A CN112132089A CN 112132089 A CN112132089 A CN 112132089A CN 202011054904 A CN202011054904 A CN 202011054904A CN 112132089 A CN112132089 A CN 112132089A
- Authority
- CN
- China
- Prior art keywords
- loss
- optical flow
- network
- convolution
- excavator
- 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
- 230000003287 optical effect Effects 0.000 title claims abstract description 27
- 238000004458 analytical method Methods 0.000 title claims abstract description 24
- 238000000034 method Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 19
- 230000006399 behavior Effects 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims abstract description 17
- 238000013135 deep learning Methods 0.000 claims abstract description 10
- 238000001514 detection method Methods 0.000 claims abstract description 3
- 230000006870 function Effects 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 7
- 238000012806 monitoring device Methods 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 235000002566 Capsicum Nutrition 0.000 claims description 3
- 239000006002 Pepper Substances 0.000 claims description 3
- 235000016761 Piper aduncum Nutrition 0.000 claims description 3
- 235000017804 Piper guineense Nutrition 0.000 claims description 3
- 235000008184 Piper nigrum Nutrition 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 150000003839 salts Chemical class 0.000 claims description 3
- 230000007423 decrease Effects 0.000 claims description 2
- 244000203593 Piper nigrum Species 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 description 10
- 230000009471 action Effects 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000009434 installation Methods 0.000 description 3
- 241000722363 Piper Species 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 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
- 230000000392 somatic effect Effects 0.000 description 1
- 230000002123 temporal effect 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
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an excavator behavior analysis method based on 3D convolution and optical flow, which comprises the following steps: s1, firstly, acquiring a data set, and performing enhanced processing on the data in the data set; s2, training a sample by using a caffe, and using a python layer as a data input layer; s3, building a network architecture of target classification based on resnet18-3D deep learning; and S4, configuring parameters of the training model and training the detection model. The excavator behavior analysis method based on the 3D convolution and the optical flow is combined with a deep learning method, the accuracy rate is improved by utilizing the time sequence of the video, and great application value can be generated.
Description
Technical Field
The invention belongs to the technical field of video monitoring analysis, and particularly relates to an excavator behavior analysis method based on 3D convolution and optical flow.
Background
At present, a monitoring camera is visible everywhere, but is artificially monitored in the monitoring process, so that a lot of manpower, material resources and financial resources are wasted, and the efficiency is low. Secondly, recent deep learning compares fire and heat, but the feasibility technology for video behavior analysis is few and few. The existing technology for analyzing the current video has the following defects:
1) most of the current production and life use traditional video monitoring systems, the video results are artificially stored and played back, and sometimes alarms cannot be generated in real time.
2) Research aiming at the traditional video analysis method is based on artificial feature extraction, wherein the process of feature extraction is complicated, and particularly the video analysis in a complex scene is difficult to realize.
3) In the research aiming at video analysis deep learning, currently, judgment is mostly made aiming at a single-frame picture in a video, time dimension in the video is difficult to fuse, and a behavior analysis technology mainly aims at analysis on actions such as human-to-human, human-to-object and the like. The result of the behavior is difficult to judge only by a single picture, resulting in low accuracy. And the accuracy rate is effectively improved by utilizing the time information in the video. In addition, an effective and fast classification method is provided for the limitation of the current computer hardware. And the running speed is improved while the accuracy is ensured.
4) Although the speed is high, the accuracy is not high, the training time is long, and convergence is difficult. TSN networks, while highly accurate, are slow. The current popular 3D convolution is fast, integrates the space and time information of the video, and balances the speed and the accuracy.
Disclosure of Invention
In view of the above, in order to overcome the above-mentioned drawbacks, the present invention aims to provide a method for analyzing excavator behavior based on 3D convolution and optical flow,
in order to achieve the purpose, the technical scheme of the invention is realized as follows:
an excavator behavior analysis method based on 3D convolution and optical flow comprises the following steps:
s1, firstly, acquiring a data set, and performing enhanced processing on the data in the data set;
s2, training a sample by using a caffe, and using a python layer as a data input layer;
s3, building a network architecture of target classification based on resnet18-3D deep learning;
and S4, configuring parameters of the training model and training the detection model.
Further, in step S1, the data set is obtained from the monitoring device, and the data set is obtained from the UCF101 and the HMDB 51.
Further, the data enhancement processing in step S1 includes brightness processing of the image, sharpening processing of the image, random salt and pepper processing, and fine adjustment of the rotation angle of the image ([ -10 °, 10 ° ]).
Further, the specific method of step S3 is as follows:
the data sets are sent to two branches, one 3D network receiving RGB format and one 3D network receiving optical flow;
the basic backbone network adopts a resnet18-3D residual 3D network, 16 frames of pictures are input into the network each time, the size of a convolution kernel is [ N multiplied by H multiplied by W multiplied by C multiplied by T ], wherein T is a time channel, and the time sequence of 16 pictures is subjected to fusion convolution;
the final loss function is the sum of the losses of the two networks, and then the total loss is sent to a SoftmaxWithLoss layer for classification.
Further, the specific method of step S4 is as follows:
the learning rate of the network is set to 0.01 at the beginning, the learning rate decreases by 0.1 per 10000 iterations, and 40000 iterations are performed totally; the Dropout values are set to 0.7 and 0.9, respectively; and selecting an SGD algorithm by an optimization function, checking the Loss value and the accuracy of the test set in the training process, and obtaining a proper classification model.
Further, the loss value is calculated as follows:
the loss in the training process is divided into two, one is a loss function of a space network, the other is a loss function of an optical flow network, and the loss function utilizes the cross entropy loss of standard classification;
Lossgeneral assembly＝LossRGB+LossOptical flow
Wherein, in LossOptical flowWherein C is the number of classes, N is the number of samples, P isiIs the probability value, P, of sample ii *Is the probability value of the true sample i, GiScore value, y, for each categoryiIs a predicted value of the ith category, LossOptical flowLoss value of optical flow, LossRGBFor RGB loss values, the final computed score probability value is calculated as follows, where k is the number of classes:
yt＝argmax(output(xj))
wherein, [ x ]1、x2...xi]Solving a maximum function for the eigenvalues of each class argmax ();
the final output result of the network is the sum x of the scores of the space network and the time networkiThen solving probability values of all categories through a SoftMax layer, wherein the maximum probability is the final prediction result yt。
Compared with the prior art, the excavator behavior analysis method based on the 3D convolution and the optical flow has the following advantages:
the excavator behavior analysis method based on the 3D convolution and the optical flow solves the problem that most of the existing video behavior analysis is manual monitoring or the traditional method is used for feature extraction. Not only is manpower wasted, but also the accuracy of the traditional method is low. If a deep learning method is combined, the accuracy is improved by utilizing the time sequence of the video, great application value can be generated, and the benefits are as follows:
1) the application field is wide. Behavioral information for many apparent actions can be monitored.
2) Aiming at the problem of cost, the camera is used for acquiring video information and recognizing video content through deep learning, extra manpower is not needed, and manpower, material resources and financial resources are greatly saved.
3) By applying the deep learning method, the convolution network has accurate extraction characteristics, high recognition rate, high 3D convolution speed and fusion time sequence, and the optical flow information ensures that the result is more accurate.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
fig. 1 is a schematic view of an application scenario of a monitoring device according to an embodiment of the present invention;
FIG. 2 is a network structure diagram of an analysis method according to an embodiment of the present invention;
fig. 3 is a flowchart of an analysis method according to an embodiment of the present invention.
Detailed Description
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like, indicate orientations or positional relationships based on those shown in the drawings, and are used only for convenience in describing the present invention and for simplicity in description, and do not indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and thus, are not to be construed as limiting the present invention. Furthermore, the terms "first", "second", etc. are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first," "second," etc. may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless otherwise specified.
In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art through specific situations.
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
The method is suitable for most scenes of monitoring equipment installation, and can automatically switch parameter configuration of the monitoring equipment according to real-time monitored image information.
The method can be applied to various fields, such as a campus corner of a school, places where few people go and go, and fighting events can be monitored. For example, the cable safety monitoring device can be applied to outdoor cable safety monitoring, and the cable is prevented from being damaged or stolen by misoperation of an excavator. Such as in prisons to prevent fighting events. For example, the early warning can be realized by monitoring the actions of falling of the old and the children.
The specific implementation method comprises the following steps:
fig. 1 is a schematic diagram of an application scenario: the indoor installation of the monitoring equipment is generally not less than 2.5 meters, and the outdoor installation is not less than 3 meters.
The camera can be placed in a crowd area or a target area and cannot be shielded.
Fig. 2 is a schematic diagram of a network structure: the network adopts a resnet18-3D structure, extracts a fixed frame number (default 16 frames) each time according to information acquired by video streaming, then respectively sends the fixed frame number to different networks for classification, and finally judges a classification result.
FIG. 3 is an algorithm flow chart: the algorithm combines 3D and optical flow network processing videos, 3D extracts spatial information and time information in the videos, then optical flow extracts somatic action information, and the two are fused, so that the result is greatly improved.
(i.) first, the dataset is obtained, the actual dataset is obtained from the monitoring device, and the data is obtained from an open source dataset on the network, such as UCF101, HMDB51, etc. And then carrying out data enhancement processing on the acquired data set, wherein the data enhancement processing comprises brightness processing of an image, sharpening processing of the image, random salt and pepper processing and fine adjustment of the rotation angle ([ -10 degrees, 10 degrees ]) of the image. Training a sample: the test sample was 8: 1. And (4) utilizing the cafe training and using the python layer as a data input layer.
(ii.) build a network architecture based on the object classification of resnet18-3D deep learning, the data sets are fed into two branches, one receiving the 3D network in RGB format and one receiving the 3D network of optical flow. The basic backbone network adopts a resnet18-3D residual 3D network, the network inputs 16 frames of pictures at each time, the size of a convolution kernel is [ N multiplied by H multiplied by W multiplied by C multiplied by T ], wherein T is a time channel, and the time sequence of 16 pictures is subjected to fusion convolution. The final loss function is the sum of the losses of the two networks, and then the total loss is sent to a SoftmaxWithLoss layer for classification.
(iii.) parameters of the training model were configured, the learning rate of the net was initially set to 0.01, with 10000 learning rate drops by 0.1 per iteration, 40000 total iterations. The Dropout values are set to 0.7 (spatial network) and 0.9 (temporal network), respectively. And (3) selecting an SGD (random gradient descent) algorithm by an optimization function, checking the Loss value and the accuracy of the test set in the training process, and obtaining a proper classification model.
(iv) initially, 16 pictures are taken from the acquisition device.
And (v.) sending the pictures into a network, sending the pictures into a trained model according to the pictures obtained in the step (iv), wherein the pictures are divided into two branches, firstly sending 16 frames of pictures into a space network for 3D convolution to obtain a probability value, secondly extracting the light stream of the 16 frames of pictures and sending the light stream of the 16 frames of pictures into a light stream network to obtain a probability value, and then adding the probability values obtained by the two branches and sending the sum to a SoftMax layer to obtain a final classification result.
(vi.) judging the target action type according to the result obtained in (v.).
The calculation method of the loss value in the step (iii) is shown as the following formula. The loss value is calculated as follows:
the loss in the training process is divided into two, one is a loss function of a space network, the other is a loss function of an optical flow network, and the loss function utilizes the cross entropy loss of standard classification;
Lossgeneral assembly＝LossRGB+LossOptical flow
Wherein, in LossOptical flowWherein C is the number of classes, N is the number of samples, P isiIs the probability value, P, of sample ii *Is the probability value of the true sample i, GiScore value, y, for each categoryiIs a predicted value of the ith category, LossOptical flowLoss value of optical flow, LossRGBFor RGB loss values, the final computed score probability value is calculated as follows, where k is the number of classes:
yt＝argmax(output(xj))
wherein, [ x ]1、x2...xi]Solving a maximum function for the eigenvalues of each class argmax ();
the final output result of the network is the sum x of the scores of the space network and the time networkiThen solving probability values of all categories through a SoftMax layer, wherein the maximum probability isFinal predicted result yt。
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (6)
1. An excavator behavior analysis method based on 3D convolution and optical flow is characterized by comprising the following steps:
s1, firstly, acquiring a data set, and performing enhanced processing on the data in the data set;
s2, training a sample by using a caffe, and using a python layer as a data input layer;
s3, building a network architecture of target classification based on resnet18-3D deep learning;
and S4, configuring parameters of the training model and training the detection model.
2. The excavator behavior analysis method based on 3D convolution and optical flow according to claim 1, characterized in that: in step S1, a data set is obtained from the monitoring device, and data is obtained from the UCF101 and the HMDB51 network open source data sets.
3. The excavator behavior analysis method based on 3D convolution and optical flow according to claim 1, wherein the data enhancement processing in step S1 includes brightness processing of an image, sharpening processing of an image, random salt and pepper processing, and fine adjustment of a rotation angle ([ -10 °, 10 ° ]) of an image.
4. The excavator behavior analysis method based on 3D convolution and optical flow according to claim 1, wherein the specific method of step S3 is as follows:
the data sets are sent to two branches, one 3D network receiving RGB format and one 3D network receiving optical flow;
the basic backbone network adopts a resnet18-3D residual 3D network, 16 frames of pictures are input into the network each time, the size of a convolution kernel is [ N multiplied by H multiplied by W multiplied by C multiplied by T ], wherein T is a time channel, and the time sequence of 16 pictures is subjected to fusion convolution;
the final loss function is the sum of the losses of the two networks, and then the total loss is sent to a SoftmaxWithLoss layer for classification.
5. The excavator behavior analysis method based on 3D convolution and optical flow according to claim 1, wherein the specific method of step S4 is as follows:
the learning rate of the network is set to 0.01 at the beginning, the learning rate decreases by 0.1 per 10000 iterations, and 40000 iterations are performed totally; the Dropout values are set to 0.7 and 0.9, respectively; and selecting an SGD algorithm by an optimization function, checking the Loss value and the accuracy of the test set in the training process, and obtaining a proper classification model.
6. The excavator behavior analysis method based on 3D convolution and optical flow according to claim 5, wherein: the loss value is calculated as follows:
the loss in the training process is divided into two, one is a loss function of a space network, the other is a loss function of an optical flow network, and the loss function utilizes the cross entropy loss of standard classification;
Lossgeneral assembly＝LossRGB+LossOptical flow
Wherein, in LossOptical flowWherein C is the number of classes, N is the number of samples, P isiIs the probability value, P, of sample ii *Is the probability value of the true sample i, GiScore value, y, for each categoryiIs a predicted value of the ith category, LossOptical flowLoss value of optical flow, LossRGBLoss value for RGBThe final computed score probability value is calculated as follows, where k is the number of categories:
yt＝argmax(output(xj))
wherein, [ x ]1、x2...xi]Solving a maximum function for the eigenvalues of each class argmax ();
the final output result of the network is the sum x of the scores of the space network and the time networkiThen solving probability values of all categories through a SoftMax layer, wherein the maximum probability is the final prediction result yt。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011054904.4A CN112132089A (en) | 2020-09-28 | 2020-09-28 | Excavator behavior analysis method based on 3D convolution and optical flow |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011054904.4A CN112132089A (en) | 2020-09-28 | 2020-09-28 | Excavator behavior analysis method based on 3D convolution and optical flow |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112132089A true CN112132089A (en) | 2020-12-25 |
Family
ID=73843304
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011054904.4A Pending CN112132089A (en) | 2020-09-28 | 2020-09-28 | Excavator behavior analysis method based on 3D convolution and optical flow |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112132089A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112597975A (en) * | 2021-02-26 | 2021-04-02 | 上海闪马智能科技有限公司 | Fire smoke and projectile detection method and system based on video |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127736A (en) * | 2016-06-15 | 2016-11-16 | 南京信必达智能技术有限公司 | One is parked detection method and processor |
| CN108225216A (en) * | 2016-12-14 | 2018-06-29 | 中国科学院深圳先进技术研究院 | Structured-light system scaling method and device, structured-light system and mobile equipment |
| CN109829443A (en) * | 2019-02-23 | 2019-05-31 | 重庆邮电大学 | Video behavior recognition methods based on image enhancement Yu 3D convolutional neural networks |
| CN109886225A (en) * | 2019-02-27 | 2019-06-14 | 浙江理工大学 | A kind of image gesture motion on-line checking and recognition methods based on deep learning |
| CN110175596A (en) * | 2019-06-04 | 2019-08-27 | 重庆邮电大学 | The micro- Expression Recognition of collaborative virtual learning environment and exchange method based on double-current convolutional neural networks |
| CN110188637A (en) * | 2019-05-17 | 2019-08-30 | 西安电子科技大学 | A kind of Activity recognition technical method based on deep learning |
| CN110705463A (en) * | 2019-09-29 | 2020-01-17 | 山东大学 | Video human behavior recognition method and system based on multi-mode double-flow 3D network |
| CN110991432A (en) * | 2020-03-03 | 2020-04-10 | 支付宝(杭州)信息技术有限公司 | Living body detection method, living body detection device, electronic equipment and living body detection system |
| CN111325155A (en) * | 2020-02-21 | 2020-06-23 | 重庆邮电大学 | Video motion recognition method based on residual difference type 3D CNN and multi-mode feature fusion strategy |
| CN111401207A (en) * | 2020-03-11 | 2020-07-10 | 福州大学 | Human body action recognition method based on MARS depth feature extraction and enhancement |
| US20200272823A1 (en) * | 2017-11-14 | 2020-08-27 | Google Llc | Weakly-Supervised Action Localization by Sparse Temporal Pooling Network |
-
2020
- 2020-09-28 CN CN202011054904.4A patent/CN112132089A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127736A (en) * | 2016-06-15 | 2016-11-16 | 南京信必达智能技术有限公司 | One is parked detection method and processor |
| CN108225216A (en) * | 2016-12-14 | 2018-06-29 | 中国科学院深圳先进技术研究院 | Structured-light system scaling method and device, structured-light system and mobile equipment |
| US20200272823A1 (en) * | 2017-11-14 | 2020-08-27 | Google Llc | Weakly-Supervised Action Localization by Sparse Temporal Pooling Network |
| CN109829443A (en) * | 2019-02-23 | 2019-05-31 | 重庆邮电大学 | Video behavior recognition methods based on image enhancement Yu 3D convolutional neural networks |
| CN109886225A (en) * | 2019-02-27 | 2019-06-14 | 浙江理工大学 | A kind of image gesture motion on-line checking and recognition methods based on deep learning |
| CN110188637A (en) * | 2019-05-17 | 2019-08-30 | 西安电子科技大学 | A kind of Activity recognition technical method based on deep learning |
| CN110175596A (en) * | 2019-06-04 | 2019-08-27 | 重庆邮电大学 | The micro- Expression Recognition of collaborative virtual learning environment and exchange method based on double-current convolutional neural networks |
| CN110705463A (en) * | 2019-09-29 | 2020-01-17 | 山东大学 | Video human behavior recognition method and system based on multi-mode double-flow 3D network |
| CN111325155A (en) * | 2020-02-21 | 2020-06-23 | 重庆邮电大学 | Video motion recognition method based on residual difference type 3D CNN and multi-mode feature fusion strategy |
| CN110991432A (en) * | 2020-03-03 | 2020-04-10 | 支付宝(杭州)信息技术有限公司 | Living body detection method, living body detection device, electronic equipment and living body detection system |
| CN111401207A (en) * | 2020-03-11 | 2020-07-10 | 福州大学 | Human body action recognition method based on MARS depth feature extraction and enhancement |
Non-Patent Citations (1)
| Title |
|---|
| 党伟超等: "基于改进双流法的井下配电室巡检行为识别", 《工矿自动化》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112597975A (en) * | 2021-02-26 | 2021-04-02 | 上海闪马智能科技有限公司 | Fire smoke and projectile detection method and system based on video |
| CN112597975B (en) * | 2021-02-26 | 2021-06-08 | 上海闪马智能科技有限公司 | Fire smoke and projectile detection method and system based on video |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3869459B1 (en) | Target object identification method and apparatus, storage medium and electronic apparatus | |
| CN110428522B (en) | Intelligent security system of wisdom new town | |
| CN104123544B (en) | Anomaly detection method and system based on video analysis | |
| WO2021047306A1 (en) | Abnormal behavior determination method and apparatus, terminal, and readable storage medium | |
| CN110516529A (en) | It is a kind of that detection method and system are fed based on deep learning image procossing | |
| JP2019512827A (en) | System and method for training an object classifier by machine learning | |
| CN109376637A (en) | Passenger number statistical system based on video monitoring image processing | |
| CN111401311A (en) | High-altitude parabolic recognition method based on image detection | |
| CN111582122B (en) | System and method for intelligently analyzing behaviors of multi-dimensional pedestrians in surveillance video | |
| CN106600888A (en) | Forest fire automatic detection method and system | |
| CN103927520A (en) | Method for detecting human face under backlighting environment | |
| CN111263114A (en) | Abnormal event alarm method and device | |
| CN106993188B (en) | A kind of HEVC compaction coding method based on plurality of human faces saliency | |
| CN107808152A (en) | Lift the method and face identification system of face recognition accuracy rate | |
| Lim et al. | Gun detection in surveillance videos using deep neural networks | |
| CN112287827A (en) | Complex environment pedestrian mask wearing detection method and system based on intelligent lamp pole | |
| Qiang et al. | Forest fire smoke detection under complex backgrounds using TRPCA and TSVB | |
| CN105139429A (en) | Fire detecting method based on flame salient picture and spatial pyramid histogram | |
| CN112257643A (en) | Smoking behavior and calling behavior identification method based on video streaming | |
| CN112464893A (en) | Congestion degree classification method in complex environment | |
| CN111126411B (en) | Abnormal behavior identification method and device | |
| Cao et al. | YOLO-SF: YOLO for fire segmentation detection | |
| CN112132089A (en) | Excavator behavior analysis method based on 3D convolution and optical flow | |
| CN113989732A (en) | Real-time monitoring method, system, equipment and readable medium based on deep learning | |
| CN113627504A (en) | Multi-mode multi-scale feature fusion target detection method based on generation of countermeasure network |
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 |
Application publication date: 20201225 |
|
| RJ01 | Rejection of invention patent application after publication |