CN113139444A - Space-time attention mask wearing real-time detection method based on MobileNet V2 - Google Patents
Space-time attention mask wearing real-time detection method based on MobileNet V2 Download PDFInfo
- Publication number
- CN113139444A CN113139444A CN202110376357.XA CN202110376357A CN113139444A CN 113139444 A CN113139444 A CN 113139444A CN 202110376357 A CN202110376357 A CN 202110376357A CN 113139444 A CN113139444 A CN 113139444A
- Authority
- CN
- China
- Prior art keywords
- time
- space
- mask
- wearing
- data
- 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
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000011897 real-time detection Methods 0.000 title claims abstract description 26
- 238000001514 detection method Methods 0.000 claims abstract description 43
- 238000000605 extraction Methods 0.000 claims abstract description 8
- 238000012795 verification Methods 0.000 claims abstract description 8
- 238000004140 cleaning Methods 0.000 claims abstract description 6
- 238000013135 deep learning Methods 0.000 claims abstract description 4
- 238000012360 testing method Methods 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000000694 effects Effects 0.000 claims description 9
- 238000012544 monitoring process Methods 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000005520 cutting process Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000002146 bilateral effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000008447 perception Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 230000001965 increasing effect Effects 0.000 description 5
- 241000700605 Viruses Species 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 208000015181 infectious disease Diseases 0.000 description 3
- 230000009193 crawling Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000002458 infectious effect Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 241000283690 Bos taurus Species 0.000 description 1
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 239000000443 aerosol Substances 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
- G06V40/165—Detection; Localisation; Normalisation using facial parts and geometric relationships
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a space-time attention mask wearing real-time detection method based on MobileNet V2, which comprises the following steps: collecting data samples of different persons wearing masks by using the Internet, and cleaning the data samples containing noise data to enable the data set to be a face image with qualified quality; expanding the data set; step three, dividing a data set; step four, after a tentorflow environment is configured on the server, deploying a MobilenetV2 target detection algorithm into a deep learning framework tentorflow; and step five, importing verification set data, utilizing a bottleneck structure in MobilenetV2 to construct an improved pyramid structure for feature extraction, adopting spatial attention to strengthen a face region for shallow features, adopting semantic features required by channel attention to strengthen high-level features to remove redundant semantic features, thus obtaining a new feature pyramid, gradually thinning the features in a gradual upsampling mode, and the like, thereby realizing an efficient face mask detection task.
Description
Technical Field
The invention relates to the field of real-time detection of mask wearing, in particular to a space-time attention mask wearing real-time detection method based on MobileNet V2.
Background
The infectious virus is mainly transmitted through spray and contact at present, aerosol transmission can occur under certain special conditions, and the mask is worn correctly under common working and living conditions, so that the daily protection requirement is met. In the epidemic period, it is very important that the ordinary public correctly wear the mask when going out. The method has the advantages that the prevention and control work of infectious viruses is fully done, the virus propagation path is effectively cut off, the spreading potential of epidemic situations is firmly restrained, the life safety and the body health of people are ensured, a large number of epidemic prevention personnel and detection points are required to be additionally arranged in public places such as communities, schools, units, dining halls, stations and the like, whether the mask is worn by the personnel entering and exiting the public places or not is detected one by one, however, the detection method needs to consume a large amount of manual resources, and the situations of missed detection and the like can occur when the flow of people is large. And because the epidemic situation is highly contagious and the complexity of public places, the defects of high infection risk, large working strength, narrow coverage, poor real-time performance and the like inevitably exist only through manual inspection, and the automatic mask wearing condition detection by a machine is urgently needed. The automatic detection of wearing the mask is realized based on computer vision, a series of problems faced by manual detection can be effectively relieved, and the method has important research significance.
In recent years, research on related algorithms specially applied to face mask wearing detection is less, and a 'Retinaface' proposed by Deng et al, a 'natural scene mask wearing detection algorithm for improving retinaFace' proposed by cattle et al, and a mask wearing detection algorithm based on improved YoloV3 proposed by Wang et al under a 'complex scene' are influenced by complex factors such as shielding, density and small scale existing in a natural scene, so that the detection effect is not ideal.
At present, some mask wearing real-time detection methods aiming at the defects of complex calculation and large memory occupation of the conventional improved depth network model exist, but the method is low in general efficiency, and the used network model cannot be effectively applied to a mobile terminal on the premise of ensuring a good effect.
Disclosure of Invention
Therefore, the invention provides a space-time attention mask wearing real-time detection method based on MobileNet V2, which can effectively solve the technical problems of low detection efficiency caused by complex computation of a depth network model and large memory occupation in the prior art.
In order to achieve the purpose, the invention provides a space-time attention mask wearing real-time detection method based on MobileNet V2, which comprises the following steps:
collecting data samples of different persons wearing masks by using the Internet, and cleaning the data samples containing noise data to enable the data set to be a face image with qualified quality;
step two, carrying out random cutting, rotation and mirror image operation on the cleaned data set so as to expand the data set;
step three, randomly disorganizing the expanded data set by taking a folder as a basic unit, and then dividing a training set, a verification set and a test set according to the proportion of 85%, 5% and 10%;
step four, configuring a tensegrow environment on a server, and installing a package required by image processing to deploy a MobilenetV2 target detection algorithm into a deep learning framework tensegrow;
step five, importing verification set data, constructing an improved pyramid structure by utilizing bottleneck in MobilenetV2 for feature extraction, adopting space attention to strengthen a face region for shallow layer features, adopting channel attention to strengthen required semantic features to remove redundant semantic features for high layer features, thus obtaining a new feature pyramid, and gradually thinning features in a gradual up-sampling mode;
step six, importing the processed training set into an algorithm frame for training to obtain final characteristics, extracting and segmenting a face frame by using a trained face detector during training, and then carrying out secondary classification on a segmentation area to judge whether the face has a mask;
and step seven, testing the pictures, the videos and the real-time monitoring data in the test set respectively.
Further, the tests are test precision and recall, and the calculation formula is as follows:
precision=TP/(TP+FP);
recall=TP/(TP+FN);
in the formula, TP represents a correct prediction positive sample, FP represents a wrong prediction positive sample, TN represents a correct prediction negative sample, and FN represents a wrong prediction negative sample;
when the calculation is completed, calculating the average precision AP and the average precision mAP, wherein the calculation formula is as follows:
further, in order to evaluate the robustness of the method, mask detection is carried out on single people, multiple people, dense people, people with shielding in special situations, and people under real-time monitoring in video and natural scenes in a mask detection system.
Furthermore, the single person detection is to facilitate observation and analysis to improve the precision and operability of the model for the face mask detection; the multi-person test is used for verifying the generalization capability of the model; the shielding under the special condition comprises a covering opening and a covering mouth; the test in the video is performed to verify the effect of the wearing detection of the mask in the natural scene by the algorithm.
Further, the step one of collecting data by using the internet refers to crawling and downloading published data on the internet.
Further, in the first step, a bilateral filter function is adopted for cleaning the data sample containing the noise data.
Further, the principle of dividing the proportion in the third step is to divide by using a leave-out method.
Further, the feature extraction in the fifth step is dependent on convolution operation, and the concept of the local perception area is used to fuse the spatial information and the channel information to extract the features containing information.
Further, in the sixth step, the trained face detector is used for extracting the face frame and the face frame is segmented by adopting segmentation based on the anchor point; the second category includes wearing and not wearing masks.
Further, the step seven of performing the test employs inputting the test sample into the trained network for performing the test.
Compared with the prior art, the method has the advantages that aiming at the defects of complex calculation and large memory occupation of the conventional improved deep network model, the lightweight network published in recent years is researched, and is inspired by the MobileNet proposed by Google, the mobilenetV2 is used as a main network, and partial improvement is carried out on the basis of the main network to realize an efficient face mask detection task. The improved network model can be effectively applied to a mobile terminal, and meanwhile, a good effect is guaranteed.
Further, in view of the disadvantage of poor flexibility of the existing monitoring technology, the present invention uses a movable camera to detect and classify the targets. The movable camera has high flexibility, can monitor the target in real time under a large-scale scene, responds, and effectively solves the problems of poor flexibility, small monitoring area and the like of the fixed camera.
Drawings
FIG. 1 is a schematic flow chart of a space-time attention mask wearing real-time detection method based on MobileNet V2 according to the present invention;
FIG. 2 is a schematic diagram of a real-time detection network for wearing a spatiotemporal attention mask according to the present invention based on MobileNet V2;
FIG. 3 is a diagram of an inversed Residual Block convolution process of the invention, MobileNet V2;
FIG. 4 is a high-definition image after cleaning of a space-time attention mask wearing real-time detection method based on MobileNet V2 according to the present invention;
FIG. 5 is a diagram of a single person mask detection result of the space-time attention mask wearing real-time detection method based on MobileNet V2 according to the present invention;
FIG. 6 is a diagram showing the detection result of a multi-person mask by the space-time attention mask wearing real-time detection method based on MobileNet V2;
FIG. 7 is a mask detection result diagram of a special case of the space-time attention mask wearing real-time detection method based on MobileNet V2 according to the present invention;
FIG. 8 is a video mask detection result diagram in a natural scene of the space-time attention mask wearing real-time detection method based on MobileNet V2 according to the present invention;
fig. 9 is a diagram of mask detection results in real-time monitoring in a natural scene of the space-time attention mask wearing real-time detection method based on MobileNetV 2.
Detailed Description
In order that the objects and advantages of the invention will be more clearly understood, the invention is further described below with reference to examples; it should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only for explaining the technical principle of the present invention, and do not limit the scope of the present invention.
It should be noted that in the description of the present invention, the terms of direction or positional relationship indicated by the terms "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings, which are only for convenience of description, and do not indicate or imply that the device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention.
Furthermore, it should be noted that, in the description of the present invention, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; 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 meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
Referring to fig. 1-9, the present invention provides a space-time attention mask wearing real-time detection method based on MobileNetV2, comprising:
collecting data samples of different persons wearing masks by using the Internet, and cleaning the data samples containing noise data to enable the data set to be a face image with qualified quality;
step two, carrying out random cutting, rotation and mirror image operation on the cleaned data set so as to expand the data set;
step three, randomly disorganizing the expanded data set by taking a folder as a basic unit, and then dividing a training set, a verification set and a test set according to the proportion of 85%, 5% and 10%;
step four, configuring a tensegrow environment on a server, and installing a package required by image processing to deploy a MobilenetV2 target detection algorithm into a deep learning framework tensegrow;
step five, importing verification set data, constructing an improved pyramid structure by utilizing bottleneck in MobilenetV2 for feature extraction, adopting space attention to strengthen a face region for shallow layer features, adopting channel attention to strengthen required semantic features to remove redundant semantic features for high layer features, thus obtaining a new feature pyramid, and gradually thinning features in a gradual up-sampling mode;
step six, importing the processed training set into an algorithm frame for training to obtain final characteristics, extracting and segmenting a face frame by using a trained face detector during training, and then carrying out secondary classification on a segmentation area to judge whether the face has a mask;
and step seven, testing the pictures, the videos and the real-time monitoring data in the test set respectively.
In the embodiment of the invention, after the data set is expanded, the diversity of the sample can be enhanced, so that the model can keep higher generalization capability to prevent overfitting in training; the face with qualified quality means that the proportion of the face in the image is proper. The random cutting, rotation and mirror image operation of the cleaned data set are operated in codes, which is a conventional technical means in the field; the expanding the data set refers to enhancing the data; the purpose of random scrambling is to make data present certain randomness; the training set is used for learning the network, the verification set is used for finding out the best model parameter for learning, and the test set is used for checking the performance of the algorithm; the channel attention strengthening adopted for the high-rise features is an attention mechanism; the removing of redundant semantic features refers to increasing useful feature weights and decreasing useless feature weights in an attention mechanism; the judgment of whether the face has the mask can be directly obtained based on the algorithm of the invention.
Specifically, the tests are test precision and recall, and the calculation formula is as follows:
precision=TP/(TP+FP);
recall=TP/(TP+FN);
in the formula, TP represents a correct prediction positive sample, FP represents a wrong prediction positive sample, TN represents a correct prediction negative sample, and FN represents a wrong prediction negative sample;
when the calculation is completed, calculating the average precision AP and the average precision mAP, wherein the calculation formula is as follows:
the accuracy rate in the embodiment of the present invention indicates a ratio of a true positive sample in a sample predicted as a positive sample, the recall rate indicates a ratio of a positive sample successfully predicted as a positive sample, and the Average accuracy ap (Average precision), and the Average accuracy Average map (mean Average precision) indicates an Average of Average accuracies of all categories, which reflects an overall target detection effect.
Specifically, in order to evaluate the robustness of the method, mask detection is performed on single people, multiple people, dense people, people with occlusion under special conditions, and people under real-time monitoring in video and natural scenes in a mask detection system.
Specifically, the single person detection is to facilitate observation and analysis to improve the precision and operability of the model for face mask detection; the multi-person test is used for verifying the generalization capability of the model; the shielding under the special condition comprises a covering opening and a covering mouth; the test in the video is performed to verify the effect of the wearing detection of the mask in the natural scene by the algorithm.
Specifically, the step one of collecting data by using the internet refers to crawling and downloading published data on the internet.
Specifically, in the first step, a bilateral filter function is adopted to clean the data samples containing the noise data.
Specifically, the principle of dividing the ratio in the third step is to divide by using a leave-out method.
Specifically, the feature extraction in the fifth step is a convolution operation, and extracts features including information by fusing spatial information and channel information using the concept of the local perceptual area.
Specifically, in the sixth step, the trained face detector is used for extracting the face frame and the face frame is segmented by adopting segmentation based on the anchor point; the second category includes wearing and not wearing masks.
Specifically, the step seven of performing the test uses inputting the test sample into the trained network for performing the test.
Example 1
In order to verify real-time detection, the invention adopts an external camera to carry out real-time test, and obtains very good detection effect (as shown in figure 9).
Example 2
In order to verify the generalization of the model, the method is compared with the existing face mask detection algorithm, and as can be seen from table 1, for the target detection of the face mask, the algorithm of the invention achieves an obvious detection effect, and compared with the RetinaFace algorithm and the Attention-RetinaFace, the AP values are respectively increased by 21.3% and 13.1%, and the mAP values are respectively increased by 12.3% and 16.5%;
compared with YoloV3 and improved YoloV3, the AP values were increased by 17.3% and 2.4%, respectively, while the maps were increased by 15.1% and 4%, respectively. But for the detection of the human face target, the algorithm also obtains better results.
Table 1 experimental comparison data
So far, the technical solutions of the present invention have been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of the present invention is obviously not limited to these specific embodiments. Equivalent changes or substitutions of related technical features can be made by those skilled in the art without departing from the principle of the invention, and the technical scheme after the changes or substitutions can fall into the protection scope of the invention.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention; various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A space-time attention mask wearing real-time detection method based on MobileNet V2 is characterized by comprising the following steps:
collecting data samples of different persons wearing masks by using the Internet, and cleaning the data samples containing noise data to enable the data set to be a face image with qualified quality;
step two, carrying out random cutting, rotation and mirror image operation on the cleaned data set so as to expand the data set;
step three, randomly disorganizing the expanded data set by taking a folder as a basic unit, and then dividing a training set, a verification set and a test set according to the proportion of 85%, 5% and 10%;
step four, configuring a tensegrow environment on a server, and installing a package required by image processing to deploy a MobilenetV2 target detection algorithm into a deep learning framework tensegrow;
step five, importing verification set data, constructing an improved pyramid structure by utilizing bottleneck in MobilenetV2 for feature extraction, adopting space attention to strengthen a face region for shallow layer features, adopting channel attention to strengthen required semantic features to remove redundant semantic features for high layer features, thus obtaining a new feature pyramid, and gradually thinning features in a gradual up-sampling mode;
step six, importing the processed training set into an algorithm frame for training to obtain final characteristics, extracting and segmenting a face frame by using a trained face detector during training, and then carrying out secondary classification on a segmentation area to judge whether the face has a mask;
and step seven, testing the pictures, the videos and the real-time monitoring data in the test set respectively.
2. The space-time attention mask wearing real-time detection method based on the MobileNetV2 as claimed in claim 1, wherein the tests are precision and recall, and the calculation formula is as follows:
precision=TP/(TP+FP);
recall=TP/(TP+FN);
in the formula, TP represents a correct prediction positive sample, FP represents a wrong prediction positive sample, TN represents a correct prediction negative sample, and FN represents a wrong prediction negative sample;
when the calculation is completed, calculating the average precision AP and the average precision mAP, wherein the calculation formula is as follows:
3. the space-time attention mask wearing real-time detection method based on MobileNetV2 as claimed in claim 1, wherein in order to evaluate robustness of the method of the present invention, mask detection is performed in a mask detection system for single person, multiple persons, and dense persons, persons under occlusion in special cases, persons under real-time monitoring in video and natural scenes.
4. The space-time attention mask wearing real-time detection method based on MobileNetV2 as claimed in claim 3, wherein the single person detection is to improve the accuracy and operability of the model for face mask detection for easy observation and analysis; the multi-person test is used for verifying the generalization capability of the model; the shielding under the special condition comprises a covering opening and a covering mouth; the test in the video is performed to verify the effect of the wearing detection of the mask in the natural scene by the algorithm.
5. The method for detecting the wearing of the space-time attention mask based on the MobileNetV2 as claimed in claim 1, wherein the step one comprises the step of collecting published data by using the internet.
6. The space-time attention mask wearing real-time detection method based on the MobileNetV2 as claimed in claim 1, wherein the first step is to apply a bilateral filter function to clean data samples containing noise data.
7. The space-time attention mask wearing real-time detection method based on the MobileNet V2 as claimed in claim 1, wherein the division ratio in the third step is divided by a leave-out method.
8. The space-time attention mask wearing real-time detection method based on MobileNetV2 as claimed in claim 1, wherein the feature extraction in the fifth step is a convolution operation, and the concept of local perception area is used to extract features containing information by fusing spatial information and channel information.
9. The space-time attention mask wearing real-time detection method based on MobileNetV2 as claimed in claim 1, wherein the extraction of the face frame and the segmentation using the trained face detector in the sixth step are based on anchor point segmentation; the second category includes wearing and not wearing masks.
10. The method for detecting the wearing time of the space-time attention mask based on the MobileNet V2 as claimed in claim 1, wherein the step seven of the test is performed by inputting test samples into a trained network.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110376357.XA CN113139444A (en) | 2021-04-06 | 2021-04-06 | Space-time attention mask wearing real-time detection method based on MobileNet V2 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110376357.XA CN113139444A (en) | 2021-04-06 | 2021-04-06 | Space-time attention mask wearing real-time detection method based on MobileNet V2 |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113139444A true CN113139444A (en) | 2021-07-20 |
Family
ID=76810610
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110376357.XA Pending CN113139444A (en) | 2021-04-06 | 2021-04-06 | Space-time attention mask wearing real-time detection method based on MobileNet V2 |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113139444A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114663912A (en) * | 2022-02-25 | 2022-06-24 | 青岛图灵科技有限公司 | Method and device for intelligently detecting whether dressing of police is standard, electronic equipment and storage medium |
CN114782931A (en) * | 2022-04-22 | 2022-07-22 | 电子科技大学 | Driving behavior classification method for improved MobileNetv2 network |
CN115600643A (en) * | 2022-10-17 | 2023-01-13 | 中国科学技术大学(Cn) | Method and system for rapidly predicting toxic gas |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108364023A (en) * | 2018-02-11 | 2018-08-03 | 北京达佳互联信息技术有限公司 | Image-recognizing method based on attention model and system |
CN110188817A (en) * | 2019-05-28 | 2019-08-30 | 厦门大学 | A kind of real-time high-performance street view image semantic segmentation method based on deep learning |
CN110363137A (en) * | 2019-07-12 | 2019-10-22 | 创新奇智(广州)科技有限公司 | Face datection Optimized model, method, system and its electronic equipment |
CN110516529A (en) * | 2019-07-09 | 2019-11-29 | 杭州电子科技大学 | It is a kind of that detection method and system are fed based on deep learning image procossing |
CN111680637A (en) * | 2020-06-10 | 2020-09-18 | 深延科技(北京)有限公司 | Mask detection method and detection system based on deep learning and image recognition technology |
-
2021
- 2021-04-06 CN CN202110376357.XA patent/CN113139444A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108364023A (en) * | 2018-02-11 | 2018-08-03 | 北京达佳互联信息技术有限公司 | Image-recognizing method based on attention model and system |
CN110188817A (en) * | 2019-05-28 | 2019-08-30 | 厦门大学 | A kind of real-time high-performance street view image semantic segmentation method based on deep learning |
CN110516529A (en) * | 2019-07-09 | 2019-11-29 | 杭州电子科技大学 | It is a kind of that detection method and system are fed based on deep learning image procossing |
CN110363137A (en) * | 2019-07-12 | 2019-10-22 | 创新奇智(广州)科技有限公司 | Face datection Optimized model, method, system and its electronic equipment |
CN111680637A (en) * | 2020-06-10 | 2020-09-18 | 深延科技(北京)有限公司 | Mask detection method and detection system based on deep learning and image recognition technology |
Non-Patent Citations (1)
Title |
---|
牛作东,覃涛,李捍东,陈进军: "改进 RetinaFace的自然场景口罩佩戴检测算法", 《计算机工程与应用》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114663912A (en) * | 2022-02-25 | 2022-06-24 | 青岛图灵科技有限公司 | Method and device for intelligently detecting whether dressing of police is standard, electronic equipment and storage medium |
CN114782931A (en) * | 2022-04-22 | 2022-07-22 | 电子科技大学 | Driving behavior classification method for improved MobileNetv2 network |
CN114782931B (en) * | 2022-04-22 | 2023-09-29 | 电子科技大学 | Driving behavior classification method for improving mobilenet v2 network |
CN115600643A (en) * | 2022-10-17 | 2023-01-13 | 中国科学技术大学(Cn) | Method and system for rapidly predicting toxic gas |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113139444A (en) | Space-time attention mask wearing real-time detection method based on MobileNet V2 | |
CN109819208B (en) | Intensive population security monitoring management method based on artificial intelligence dynamic monitoring | |
CN108090458B (en) | Human body falling detection method and device | |
CN111881726B (en) | Living body detection method and device and storage medium | |
CN112085010A (en) | Mask detection and deployment system and method based on image recognition | |
CN109858371A (en) | The method and device of recognition of face | |
CN111091110A (en) | Wearing identification method of reflective vest based on artificial intelligence | |
CN105022999A (en) | Man code company real-time acquisition system | |
CN112115775A (en) | Smoking behavior detection method based on computer vision in monitoring scene | |
CN109829997A (en) | Staff attendance method and system | |
RU2713876C1 (en) | Method and system for detecting alarm events when interacting with self-service device | |
CN113657150A (en) | Fall detection method and device and computer readable storage medium | |
CN112257643A (en) | Smoking behavior and calling behavior identification method based on video streaming | |
Nadeem et al. | A survey of artificial intelligence and internet of things (IoT) based approaches against COVID-19 | |
Zhu et al. | Towards automatic wild animal detection in low quality camera-trap images using two-channeled perceiving residual pyramid networks | |
CN113887318A (en) | Embedded power violation detection method and system based on edge calculation | |
Chachere et al. | Real Time Face Mask Detection by using CNN | |
CN113506274B (en) | Detection system for human cognitive condition based on visual saliency difference map | |
CN103986882A (en) | Method for image classification, transmission and processing in real-time monitoring system | |
Gupta et al. | Multilevel Face Mask Detection System using Ensemble based Convolution Neural Network | |
CN109684990A (en) | A kind of behavioral value method of making a phone call based on video | |
CN113052139A (en) | Deep learning double-flow network-based climbing behavior detection method and system | |
Limbasiya et al. | COVID-19 face mask and social distancing detector using machine learning | |
CN115049875A (en) | Detection method for wearing insulating gloves in transformer substation based on deep learning | |
CN106846527B (en) | A kind of attendance checking system based on recognition of face |
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: 20210720 |
|
RJ01 | Rejection of invention patent application after publication |