CN114187280A - Method and device for detecting iron receiving state - Google Patents
Method and device for detecting iron receiving state Download PDFInfo
- Publication number
- CN114187280A CN114187280A CN202111531771.XA CN202111531771A CN114187280A CN 114187280 A CN114187280 A CN 114187280A CN 202111531771 A CN202111531771 A CN 202111531771A CN 114187280 A CN114187280 A CN 114187280A
- Authority
- CN
- China
- Prior art keywords
- iron
- taphole
- image
- frame
- module
- 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
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 title claims abstract description 520
- 229910052742 iron Inorganic materials 0.000 title claims abstract description 260
- 238000000034 method Methods 0.000 title claims abstract description 112
- 238000001514 detection method Methods 0.000 claims abstract description 151
- 238000000605 extraction Methods 0.000 claims abstract description 23
- CWYNVVGOOAEACU-UHFFFAOYSA-N Fe2+ Chemical compound [Fe+2] CWYNVVGOOAEACU-UHFFFAOYSA-N 0.000 claims abstract description 9
- 230000002708 enhancing effect Effects 0.000 claims abstract description 6
- 230000008569 process Effects 0.000 claims description 45
- 238000004590 computer program Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 17
- 238000009529 body temperature measurement Methods 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 11
- 238000005259 measurement Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 description 15
- 230000006870 function Effects 0.000 description 7
- 238000011897 real-time detection Methods 0.000 description 5
- 230000007547 defect Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000003672 processing method Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011076 safety test Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 229910000805 Pig iron Inorganic materials 0.000 description 1
- 238000003723 Smelting Methods 0.000 description 1
- 229910000746 Structural steel Inorganic materials 0.000 description 1
- 238000010924 continuous production Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000010079 rubber tapping Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30136—Metal
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
Some embodiments of the present application provide a method and apparatus for detecting a ferrous state, the method comprising: acquiring multi-channel taphole video stream data; decoding the multi-channel taphole video stream data to obtain a plurality of frame taphole images, wherein each taphole corresponds to at least one frame of taphole image; inputting the multi-frame taphole image into a target identification detection network, and obtaining the iron flow detection information of at least part of the taphole through the target identification detection network, wherein the target identification detection network comprises a multi-scale feature extraction module, a channel attention module and a prediction regression module, the multi-scale feature extraction module is configured to obtain feature maps of different scales of the same taphole based on the input taphole image, the channel attention module is used for enhancing the channel attention of the feature maps, and the prediction regression module is configured to obtain the iron flow detection information of at least part of the taphole according to the feature maps output by the channel attention module.
Description
Technical Field
The application relates to the field of detection of a receiving state, in particular to a method and a device for detecting the receiving state.
Background
Blast furnace smelting is a continuous process for reducing iron ore to pig iron, and molten iron is a main product of a blast furnace. When tapping a blast furnace, generally, iron cans (also called torpedo cans and ladles) are conveyed to the lower part of a corresponding iron notch, then molten iron is contained in the iron cans, and each iron can receives iron once from the beginning to the end of containing the molten iron. The real-time detection of the state of the iron flow in the iron receiving process is particularly important, and the detection contents mainly comprise iron flow detection, iron receiving safety detection, iron flow temperature detection and the like.
The existing detection method is shown in fig. 1: for the detection of the iron flow, a traditional image processing method is adopted, firstly, an iron flow video frame is obtained, then image preprocessing is carried out, and Hough transform, feature detection and image segmentation are carried out, so that the iron flow is identified. For the temperature detection of the iron flow, the manual temperature measurement is mainly carried out by manually collecting the molten iron, or the whole iron receiving opening is shot by using a thermal infrared imager to carry out real-time temperature measurement on the molten iron. For the safety detection of the iron receiving, the judgment is mainly carried out by manually observing the monitoring video of the iron receiving, and whether a large amount of molten iron is sprayed out of the tank car is observed.
It is easy to understand that, for the detection of the iron flow, the traditional image processing method is adopted, the environment factors greatly affect the detection effect of the iron flow, and the robustness of the detection algorithm is too low; because the real-time detection of the iron flows is required to be carried out on a plurality of iron openings at the same time, if each iron opening respectively and independently obtains the video flow for detection, the efficiency of the whole detection system is too low; for the temperature detection of the iron flow, manual collection is too complicated, and the measurement process is discontinuous. The thermal infrared imager is adopted for detection, a fixed measurement area is usually set in a shooting range, and if the fluctuation of the iron flow is large, the temperature of the iron flow cannot be measured. For the safety detection of iron receiving, the manual observation mode judges whether a large amount of iron flow is ejected out of the tank car through human eyes, the efficiency is too low, only spot check can be carried out, and real-time detection cannot be carried out.
Therefore, how to improve the existing iron-bearing state detection effect becomes a technical problem to be solved urgently.
Disclosure of Invention
Embodiments of the present application provide a method and an apparatus for detecting a state of a receiving object, where in some embodiments of the present application, a GPU decoding framework is used to perform image frame processing on multiple video streams, so as to greatly improve image acquisition efficiency. In some embodiments of the application, the iron flow and the iron notch of the tank car are detected and segmented based on a deep learning method, the algorithm robustness is high, and the detection precision is high. In some embodiments of the application, the method of fusion and registration of the visible light and the infrared image is used, so that the accuracy of iron flow temperature measurement is improved, and meanwhile, the detection effect on the state with large iron flow fluctuation is good. In some embodiments of the application, the safety state of the received iron is monitored in real time based on an image detection method, and when a large amount of molten iron overflows, early warning can be timely realized.
In a first aspect, some embodiments of the present application provide a method of detecting an iron-bearing state, the method comprising: acquiring multi-channel taphole video stream data; decoding the multi-channel taphole video stream data to obtain a plurality of frame taphole images, wherein each taphole corresponds to at least one frame of taphole image; the method comprises the steps of inputting the multi-frame iron notch image into a target identification detection network, and obtaining iron flow detection information of at least part of iron notches through the target identification detection network, wherein the target identification detection network comprises a multi-scale feature extraction module, a channel attention module and a prediction regression module which are sequentially connected, the multi-scale feature extraction module is configured to obtain feature maps of the same iron notch in different scales based on the input iron notch image, the channel attention module is used for enhancing the channel attention of the feature maps, the prediction regression module is configured to obtain the iron flow detection information of at least part of iron notches according to the feature maps output by the channel attention module, and the iron flow detection information comprises an iron flow target frame and an iron flow mask obtained through identification.
Some embodiments of the application can realize good identification and detection on the iron flow of different scales at each stage of iron receiving, and some embodiments of the application add an attention mechanism, enhance the channel attention of the characteristic diagram, and finally improve the accuracy of the obtained iron flow detection information. Some embodiments of the present application input multiple frames of images of multiple tapholes into one target detection and identification network at the same time, and then output the detection results of all the tapholes.
In some embodiments, the method further comprises: obtaining tank truck detection information through the target identification detection network; setting a safe iron receiving area according to the detection information of the tank opening vehicle; and obtaining a safety detection result of the iron receiving process by confirming whether the iron flow target box is positioned in the safe iron receiving area.
Some embodiments of this application detect information and jar mouthful car detection information real time monitoring receiving the safe state of iron through detecting the iron current simultaneously, when the molten iron spills over in a large number, can realize timely early warning.
In some embodiments, the obtaining a safety detection result of the ironed process by confirming whether the iron flow target box is located in the safe ironed area includes: if the iron flow target frame is confirmed to be located in the safe iron receiving area, judging that the iron receiving process is safe; and if at least part of the iron flow target frame is confirmed to be positioned outside the safe iron receiving area, judging that the iron flow has overflow.
According to some embodiments of the application, the iron flow and the tank car mouth are separated through the target identification detection network, the molten iron overflow area is set according to the detected tank car mouth, the contact ratio (IOU) of the iron flow (namely iron flow detection information) and the overflow area is calculated, whether a large amount of molten iron overflows in the iron receiving process is judged through setting a reasonable threshold value, and the problem that the detection efficiency is low and real-time detection cannot be carried out due to the manual observation mode is effectively solved.
In some embodiments, the method further comprises: acquiring multi-channel taphole infrared video stream data; extracting a plurality of frames of infrared images from the infrared video stream data; and acquiring a temperature measurement result of the received iron according to the multi-frame infrared image and the iron flow detection information.
According to some embodiments of the application, the iron flow temperature is comprehensively measured through the infrared image and the visible light image, and the accuracy of the temperature value obtained through measurement is improved.
In some embodiments, the obtaining a measured iron temperature measurement result according to the plurality of frames of infrared images and the iron flow detection information includes: acquiring image scale factors of an infrared image and a visible light image, wherein the visible light image is obtained by analyzing the multi-channel taphole video stream data; acquiring the pixel offset of any frame of infrared image and the visible light image of the corresponding frame; and superposing the iron flow detection information to the infrared image according to the image scale factor and the pixel offset to finish temperature measurement.
According to some embodiments of the application, the mapping relation between the infrared image and the visible light image is established by obtaining the image scale factor and the pixel offset, and finally the accuracy of temperature measurement is improved.
In some embodiments, the acquiring image scale factors of the infrared image and the visible light image comprises: and calculating the pixel difference between the infrared image and the visible light image between every two circle centers of the calibration plate by adopting a calibration method to obtain the scaling of the image, wherein the scaling of the image is used for representing the image scale factor.
Some embodiments of the application adopt a calibration method, calculate the pixel difference between the infrared and visible light images between two centers of circles of the calibration plate, obtain the scaling of the images, and unify the sizes of the space objects on the two images.
In some embodiments, the obtaining a pixel offset of any frame of infrared image from a corresponding frame of visible light image includes: and calculating the pixel difference of corresponding points according to the pixel coordinate positions of the circular coordinate positions in the calibration plate in the infrared image and the visible light image to obtain the pixel offset.
Some embodiments of the application need to determine the pixel offset to complete the alignment operation of the infrared image and the corresponding visible light image, and finally achieve that the iron flow detection information is superimposed on the infrared image.
In some embodiments, the decoding the multiple channels of taphole video stream data to obtain multiple frames of taphole images includes: and decoding the multi-channel taphole video stream data through a GPU decoding frame to obtain the multi-frame taphole image.
Some embodiments of the application use the GPU decoding framework to perform image frame processing on multiple paths of video streams, thereby greatly improving the image acquisition efficiency.
In some embodiments, the target identification detection network is obtained by training an identification detection network, wherein in the training process, a mean and a variance of the same batch of data are solved, and then normalization operation is performed, and the mean and the variance used in the inference stage are expected values of the mean and the variance of the whole training sample, which are used as the mean and the variance of the BN when we perform inference.
Some embodiments of the present application improve the performance of the resulting network model through this training method.
In some embodiments of the present application, the channel attention module determines the importance of each pixel based on a scale factor in the batch normalization.
Different from a method for measuring the importance of pixels by using a pooling method in the related art, the embodiment of the application applies the scale factors in batch normalization to the spatial dimension to measure the importance of the pixels, and the result accuracy is improved.
In a second aspect, some embodiments of the present application provide an apparatus for detecting a ferrous state, the apparatus comprising: the multi-channel taphole video stream data acquisition module is configured to acquire multi-channel taphole video stream data; the decoding module is configured to decode the multi-channel taphole video stream data to obtain a plurality of frame taphole images, wherein each taphole corresponds to at least one frame of taphole image; the iron flow detection information acquisition module is configured to input the multiple frames of iron notch images into a target identification detection network, and obtain iron flow detection information of at least part of iron notches through the target identification detection network, wherein the target identification detection network comprises a multi-scale feature extraction module, a channel attention module and a prediction regression module which are sequentially connected, the multi-scale feature extraction module is configured to acquire feature maps of the same iron notch in different scales based on the input iron notch images, the channel attention module is used for enhancing the channel attention of the feature maps, the prediction regression module is configured to acquire the iron flow detection information of at least part of iron notches according to the feature maps output by the channel attention module, and the iron flow detection information comprises an identified iron flow target frame and an iron flow mask.
In a third aspect, some embodiments of the present application provide a computer program product comprising a computer program, wherein the computer program, when executed by a processor, is adapted to implement the method according to any of the embodiments of the first aspect.
In a fourth aspect, some embodiments of the present application provide a computer-readable storage medium on which a computer program is stored, which when executed by a processor, may implement the method according to any of the embodiments of the first aspect.
In a fifth aspect, some embodiments of the present application provide an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the program, may implement the method according to any of the embodiments of the first aspect.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
FIG. 1 is a flowchart of a method for detecting a state of iron according to a related embodiment;
FIG. 2 is an architecture diagram of an identification detection network according to an embodiment of the present application;
FIG. 3 is a flowchart of a method for detecting a state of iron according to an embodiment of the present disclosure;
FIG. 4 is a second architecture diagram of an identification detection network according to an embodiment of the present application;
FIG. 5 is an architecture diagram of a aisle attention module provided in an embodiment of the present application;
FIG. 6 is one of the relationship diagrams between the reasonable iron-receiving area and the iron flow target box set according to the embodiment of the present application;
fig. 7 is a second flowchart of a method for detecting a receiving state according to an embodiment of the present application;
FIG. 8 is a block diagram illustrating an apparatus for detecting a state of a received iron according to an embodiment of the present disclosure;
fig. 9 is a schematic composition diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.
At least in order to overcome many problems existing in the background art, some embodiments of the present application need to acquire multiple paths of video stream data for detecting a bitstream, decode a video frame by using a GPU decoding frame, and input the video frame to a target identification detection network, thereby greatly improving the efficiency of image acquisition and solving the disadvantages of the existing solutions. Some embodiments of the application design a target detector based on a convolutional neural network for the target identification and detection of the iron flow, and use a deep learning and traditional image processing method to detect and segment the iron flow and the iron notch of the tank car, so that the robustness is greatly improved, and the defects of the existing scheme are overcome. In some embodiments of the application, for the measurement of the iron flow temperature, a method of fusing an infrared image and a visible light image is adopted, an infrared image frame and a visible light image frame are simultaneously obtained, then image registration is performed, and the temperature of the iron flow is obtained from the registered image according to the iron flow information identified in the visible light image, so that if the iron flow fluctuates greatly, the iron flow can be well identified in the visible light, the temperature of the iron flow can be more accurately obtained, and the defects of the existing scheme are overcome. In other embodiments of the application, in order to monitor whether molten iron overflows in a large amount in the process of receiving iron in real time, detection information of iron flow and detection information of a tank wagon mouth are fused for safety detection. For example, the iron flow and the tank car mouth are divided by the identification and detection network, the molten iron overflow area is set according to the detected tank car mouth, the contact ratio (IOU) of the iron flow and the overflow area is calculated, and whether a large amount of molten iron overflows in the iron receiving process is judged by setting a reasonable threshold value, so that various defects existing in the existing scheme are overcome.
Referring to fig. 2, some embodiments of the present application provide an identification detection network, which may be trained to obtain a target identification detection network, and the target identification detection network processes an input image to obtain a molten iron, a molten iron position (i.e., a molten iron target frame), and a tank opening car, a tank opening car position (i.e., a target frame corresponding to the tank opening car), and the like included in the input image. That is to say, the object recognition and detection network obtained by the recognition and detection network provided in fig. 2 has a function of example segmentation of the image, and recognizes various types of objects included in the image and positions and specific classifications of the objects, in the embodiment of the present application, the objects to be recognized on the image at least include the iron flow and the tank opening car.
The recognition detection network of fig. 2 includes a feature extraction module 110, a channel attention module 120, and a predictive regression module 130 connected in series. The feature extraction module 110 is configured to receive an input training image or an image to be segmented, and extract features of multiple scales on the image to obtain a multi-scale feature map, so that the defect that the technical scheme cannot be applied due to large size transformation of the iron flow at different stages of iron receiving can be effectively overcome. The channel attention module 120 is configured to receive the multi-scale feature map output by the feature extraction module 110 and enhance the channel attention to exclude the influence of flames and the like on the iron flow detection result. The prediction regression module 130 may output the predicted ith target classification value, the ith target frame location (e.g., vertex coordinates of the output target frame), and the ith target segmentation mask (i.e., the edge of the ith target on the image). For example, in some embodiments of the present application, the ith target includes a stream of iron and a tank car, respectively.
The process of training the recognition detection network of fig. 2 is briefly described below. Carrying out image enhancement operation on the input training picture: horizontal flipping, morphing scaling, etc. Reshape operation is carried out to the training picture of input, guarantees that the yardstick is 2^ n, and n > ═ 6, does not produce the decimal after guaranteeing to sample. During training, some embodiments of the present application solve for the mean and variance of the same batch of data, and then perform normalization. The mean and variance used in the inference stage are the expected mean and variance values of the whole training sample, and are used as the mean and variance of the Batch Normalization (Batch Normalization) BN when we perform inference.
That is to say, in some embodiments of the present application, the target identification detection network is obtained by training the identification detection network, wherein in the training process, the mean and variance of the same batch of data are solved, and then normalization operation is performed, and the mean and variance used in the inference stage are expected values of the mean and variance of the whole training sample, which are used as the mean and variance of the BN performed when we perform inference.
The following exemplarily illustrates a process of acquiring relevant parameters of the iron receiving process according to the trained target recognition detection network.
The process of obtaining the detection result of the iron flow is exemplarily set forth below with reference to fig. 3.
As shown in fig. 3, some embodiments of the present application provide a method of detecting an iron-bearing state, the method comprising: s101, acquiring multi-channel taphole video stream data; s102, decoding the multi-channel taphole video stream data to obtain a plurality of frames of taphole images, wherein each taphole corresponds to at least one frame of taphole image; and S103, inputting the multi-frame taphole image into a target identification detection network, and obtaining the iron flow detection information of at least part of the taphole through the target identification detection network, wherein the target identification detection network comprises a multi-scale feature extraction module, a channel attention module and a prediction regression module, the multi-scale feature extraction module is configured to obtain feature maps of the same taphole in different scales based on the input taphole image, the channel attention module is used for enhancing the channel attention of the feature maps, the prediction regression module is configured to obtain the iron flow detection information of at least part of the taphole according to the feature maps output by the channel attention module, and the iron flow detection information comprises an identified iron flow target frame and an iron flow mask.
It is understood that, since the iron flow image acquisition unit is provided for each of the plurality of iron openings, the iron openings are processed, so that the iron flow detection information of the plurality of iron openings can be acquired simultaneously. In addition, the multi-scale feature extraction module of the target identification and detection network model realizes good identification and detection of the iron flow with different scales at each stage of the iron receiving process, and in some embodiments of the application, the channel attention module is introduced into the target identification and detection network model to add an attention mechanism, so that the channel attention of the feature diagram is enhanced, and the accuracy of the obtained iron flow detection information is finally improved.
The characteristics of the iron flow detection result obtained by some embodiments of the present application are briefly described below with reference to the network model of fig. 4. Fig. 4 differs from fig. 2 in that fig. 4 provides an exemplary implementation architecture for the modules of fig. 2.
As shown in fig. 4, a method of detecting an iron-receiving state according to some embodiments of the present application includes: and acquiring an iron flow image to be segmented, inputting the iron flow image to be segmented into the feature extraction network (corresponding to the feature extraction module in fig. 2) in fig. 4, and then performing feature extraction by using ResNet101-FPN as a backbone network backbone to obtain a multi-scale feature map. And then, the multi-scale feature map is subjected to channel attention enhancement through a Channel Attention Module (CAM), and then the multi-scale attention map is input into a prediction regression network to predict the object type, the regression box and the mask.
Because the target detected by some embodiments of the present application is the iron flow, and the size of the iron flow is variable in the actual iron receiving process, the iron flow is large at the initial time of iron receiving, and the iron flow is small at the time when the iron receiving is about to end, the network model of some embodiments of the present application needs to have good identification capability for both large scale and small scale, so as to ensure accurate identification of the iron flow in the whole iron receiving process. For example, the input picture is first subjected to feature extraction by using ResNet101, where ResNet101 is five parts (C1, C2, C3, C4, and C5) in fig. 4, where each part represents one convolution layer, and after four times of downsampling, multi-scale feature maps of P2, P3, P4, and P5 are obtained, and then the features are further reconstructed by using convolution of 3x3, so that the reconstructed multi-scale features P2, P3, P4, and P5 are obtained. It is understood that the feature extraction network in fig. 4 can be used for realizing good identification and detection of iron flows with different scales at each stage of iron receiving.
Because the characteristics of the iron flow are very similar to those of the flame, the characteristics of the iron flow are obviously different from those of the whole environment, and the characteristics of the similar flame have channel similarity, some embodiments of the application add an attention mechanism (namely a channel attention module) in a network model to enhance the channel attention of a characteristic diagram, so that the network selectively enhances the characteristics with the largest information amount, and the post-processing fully utilizes the characteristics and inhibits useless characteristics.
For example, the specific structure of the channel attention module, i.e., the CAM module, is shown in fig. 5, F1 represents the input feature, Mc represents the output feature, γ is the scale factor of each channel, the weight is W γ, and the scale factor of BN is applied to the spatial dimension to measure the importance of the pixel, i.e., the pixel normalization.
The channel attention module uses the scale factors in the Batch Normalization (BN) whose formula is calculated as follows, the scale factors measure the variance of the channels and indicate their importance, where μ B is the mean, δ B is the standard deviation, and γ and β are trainable affine transformation parameters.
Wherein, Bin is an input value set and takes the value of the input image pixel value. BN (bin) and BN (out) are output results. μ B is the mean value, which is the pixel mean value of the input image (i.e. averaging Bin). δ B is the standard deviation, which is taken from the pixel standard deviation of the input image (i.e. the standard deviation is calculated for Bin). ε is a constant added to prevent the denominator from being 0, and has a value of 0.00001. Gamma and beta are respectively trainable affine transformation parameters, the value taking condition is obtained by network training, and the initial value is a random value.
Inputting the multi-scale attention feature map into a prediction regression network (corresponding to a prediction regression module in fig. 1) in fig. 4, firstly setting a predetermined Region of interest (Region of interest) RoI for each point in the multi-scale feature map so as to obtain a plurality of candidate rois, filtering out a part of the candidate rois by using a Region suggestion network (RPN), then performing roiign operation on the rest rois, finally performing category and coordinate prediction through a full connection layer (FC), and performing mask (mask) prediction through a Full Convolution Network (FCN).
The following exemplary method for obtaining safety tests for iron-receiving processes.
In some embodiments of the present application, the method of detecting an iron-receiving state described in fig. 3 further comprises: a process for obtaining safety test results of a process under iron, the process illustratively comprising: obtaining tank truck detection information through the target identification detection network; setting a safe iron receiving area according to the detection information of the tank opening vehicle; and obtaining a safety detection result of the iron receiving process by confirming whether the iron flow target box (namely the iron flow target box included in the iron flow detection information obtained in the figure 3) is positioned in the safe iron receiving area. That is to say, some embodiments of this application are through detecting the safe state that the indisputable real time monitoring received of indisputable detection information and jar mouthful car detection information simultaneously, when the molten iron spills over in a large number, can realize timely early warning. For example, in some embodiments of the present application, the obtaining a safety detection result of a ironed process by confirming whether the iron flow target box is located in the safe ironed area includes: if the iron flow target frame is confirmed to be located in the safe iron receiving area, judging that the iron receiving process is safe; and if at least part of the iron flow target frame is confirmed to be positioned outside the safe iron receiving area, judging that the iron flow has overflow.
According to some embodiments of the application, the iron flow and the tank car mouth are separated through the target identification detection network, the molten iron overflow area is set according to the detected tank car mouth, the contact ratio (IOU) of the iron flow (namely iron flow detection information) and the overflow area is calculated, whether a large amount of molten iron overflows in the iron receiving process is judged through setting a reasonable threshold value, and the problem that the detection efficiency is low and real-time detection cannot be carried out due to the manual observation mode is effectively solved.
It should be noted that, a schematic diagram of the iron receiving safety detection method is shown in fig. 6, in the diagram, a dark color frame 501 is a detected iron flow target frame BBox, a light color frame 502 is a reasonable iron receiving area (i.e., a safe iron receiving area) set according to a detected tank car port, whether the iron receiving process is performed safely is determined by determining whether the BBox of the iron flow is in the iron receiving area BBox, and if the iron flow overflows in a large amount (i.e., the iron receiving process is unsafe or dangerous), an alarm is given.
The process of obtaining the temperature of the iron stream is exemplarily described below.
In some embodiments of the present application, the method for detecting a state of iron further includes a process of obtaining a temperature measurement of the iron flow, for example, the process exemplarily includes: acquiring multi-channel taphole infrared video stream data; extracting a plurality of frames of infrared images from the infrared video stream data; and obtaining a measurement result of the iron flow temperature by fusing the multi-frame infrared image and the iron flow detection information. According to some embodiments of the application, the iron flow temperature is comprehensively measured through the infrared image and the visible light image, and the accuracy of the temperature value obtained through measurement is improved.
The technical means of some embodiments of the present application need to be adopted to solve the following technical problems when fusing the infrared image and the iron flow detection information. Because the focal lengths of infrared light and visible light are different, the imaging sizes of the space object on the two images are different, meanwhile, the optical center of the used infrared/visible light hardware system has deviation in the Y direction, and even if the images are zoomed according to the focal lengths, the imaging sizes of the same object on different images cannot be the same. Some embodiments of the application adopt a calibration method, calculate the pixel difference between the infrared and visible light images between two centers of circles of the calibration plate, obtain the scaling of the images, and unify the sizes of the space objects on the two images. In addition, after the infrared and visible light images are scaled to a uniform size, the offset distance (x, y) needs to be known when the small-sized infrared image is moved to the visible light image, so that some embodiments of the present application also need to determine the pixel offset to complete the alignment operation of the infrared image and the corresponding visible light image, and finally, the iron flow detection information is superimposed on the infrared image.
That is to say, in some embodiments, the above obtaining the iron flow temperature measurement result by fusing the multiple frames of infrared images and the iron flow detection information includes: acquiring image scale factors of an infrared image and a visible light image, wherein the visible light image is obtained by analyzing the multi-channel taphole video stream data; acquiring the pixel offset of any frame of infrared image and the visible light image of the corresponding frame; and superposing the iron flow detection information to the infrared image according to the image scale factor and the pixel offset to finish temperature measurement. According to some embodiments of the application, the mapping relation between the infrared image and the visible light image is established by obtaining the image scale factor and the pixel offset, and finally the accuracy of temperature measurement is improved.
For example, the process of acquiring the image scale factor illustratively includes: and calculating the pixel difference between the infrared image and the visible light image between every two circle centers of the calibration plate by adopting a calibration method to obtain the scaling of the image, wherein the scaling of the image is used for representing the image scale factor. Some embodiments of the application adopt a calibration method, calculate the pixel difference between the infrared and visible light images between two centers of circles of the calibration plate, obtain the scaling of the images, and unify the sizes of the space objects on the two images.
For example, the process of acquiring the pixel shift amount (i.e. the pixel shift amount of the infrared image of any frame and the visible light image of the corresponding frame) exemplarily includes: and calculating the pixel difference of corresponding points according to the pixel coordinate positions of the circular coordinate positions in the calibration plate in the infrared image and the visible light image to obtain the pixel offset. Some embodiments of the application need to determine the pixel offset to complete the alignment operation of the infrared image and the corresponding visible light image, and finally achieve that the iron flow detection information is superimposed on the infrared image.
In order to improve image obtaining efficiency, in some embodiments of the present application, the decoding the multiple channels of taphole video stream data to obtain multiple frames of taphole images includes: and decoding the multi-channel taphole video stream data through a GPU decoding frame to obtain the multi-frame taphole image. That is, some embodiments of the present application use the GPU decoding framework to perform image frame processing on multiple video streams, which greatly improves image acquisition efficiency.
The process of detecting the iron receiving state of the blast furnace iron flow according to some embodiments of the present application will be described below with reference to fig. 7.
As shown in fig. 7, the ferrous state detection method in some embodiments of the present application includes: the method comprises the steps of obtaining a plurality of paths of video streams, namely obtaining a plurality of paths of taphole video streams and obtaining a plurality of paths of taphole infrared video streams, decoding the obtained plurality of paths of taphole video streams by adopting a GPU to obtain a plurality of frames of visible light images, inputting the images into a target identification detection network, and simultaneously obtaining tank car mouth detection information and iron stream detection information. And in addition, a received iron safety detection result is obtained according to the iron flow detection information and the tank car mouth detection information, and safety detection information is output. In addition, image frames are extracted from the obtained multi-channel iron notch infrared video streams to obtain infrared images of the iron notches, and the infrared images and iron notch detection information of corresponding frames are subjected to image registration and fusion to obtain iron notch temperature information. And judging whether the video stream is finished or not according to the acquired iron stream detection information, the security detection information and the iron stream temperature information, repeating the process to acquire a processing result of the next frame if the video stream is not finished, and finishing the method for detecting the iron receiving state if the video stream is finished.
Referring to fig. 8, fig. 8 shows an apparatus for detecting a state of a receiving iron provided in an embodiment of the present application, it should be understood that the apparatus corresponds to the embodiment of the method in fig. 3, and can perform various steps related to the embodiment of the method, and specific functions of the apparatus can be referred to the description above, and detailed descriptions are appropriately omitted herein to avoid repetition. The device comprises at least one software functional module which can be stored in a memory in the form of software or firmware or solidified in an operating system of the device, and the device for detecting the iron-receiving state comprises: the system comprises a multi-channel taphole video stream data acquisition module 101, a decoding module 102 and a taphole detection information acquisition module 103.
The multi-channel taphole video stream data acquisition module 101 is configured to acquire multi-channel taphole video stream data.
The decoding module 102 is configured to decode the multiple channels of taphole video stream data to obtain multiple frames of taphole images, where each taphole corresponds to at least one frame of taphole image.
The iron flow detection information acquisition module 103 is configured to input the multiple frames of iron notch images into a target identification detection network, and obtain iron flow detection information of at least part of iron notches through the target identification detection network, where the target identification detection network includes a multi-scale feature extraction module, a channel attention module, and a prediction regression module, the multi-scale feature extraction module is configured to acquire feature maps of different scales of the same iron notch based on the input iron notch images, the channel attention module is used to enhance the channel attention of the feature maps, the prediction regression module is configured to acquire iron flow detection information of at least part of iron notches according to the feature maps output by the channel attention module, and the iron flow detection information includes an identified iron flow target box and an iron flow mask.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the apparatus described above may refer to the corresponding process in the foregoing method, and will not be described in too much detail herein.
Some embodiments of the present application provide a computer program product comprising a computer program, wherein the computer program, when executed by a processor, is adapted to implement any of the above-described methods of detecting a ferrous state.
Some embodiments of the present application provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, may implement any of the above-described methods of detecting a ferrous state.
As shown in fig. 9, some embodiments of the present application provide an electronic device 500, where the electronic device 500 includes a memory 510, a processor 520, and a computer program stored on the memory 510 and executable on the processor 520, where the processor 520 may implement any one of the above-described methods for detecting a ferrous state when reading the program from the memory 510 through a bus 530 and executing the program.
Processor 520 may process digital signals and may include various computing structures. Such as a complex instruction set computer architecture, a structurally reduced instruction set computer architecture, or an architecture that implements a combination of instruction sets. In some examples, processor 520 may be a microprocessor.
Memory 510 may be used to store instructions that are executed by processor 520 or data related to the execution of the instructions. The instructions and/or data may include code for performing some or all of the functions of one or more of the modules described in embodiments of the application. The processor 520 of the disclosed embodiments may be used to execute instructions in the memory 510 to implement the method shown in fig. 3. Memory 510 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memory known to those skilled in the art.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative, and for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Claims (14)
1. A method of detecting an iron-bearing condition, the method comprising:
acquiring multi-channel taphole video stream data;
decoding the multi-channel taphole video stream data to obtain a plurality of frame taphole images, wherein each taphole corresponds to at least one frame of taphole image;
the method comprises the steps of inputting the multi-frame iron notch image into a target identification detection network, and obtaining iron flow detection information of at least part of iron notches through the target identification detection network, wherein the target identification detection network comprises a multi-scale feature extraction module, a channel attention module and a prediction regression module which are sequentially connected, the multi-scale feature extraction module is configured to obtain feature maps of the same iron notch in different scales based on the input iron notch image, the channel attention module is used for enhancing the channel attention of the feature maps, the prediction regression module is configured to obtain the iron flow detection information of at least part of iron notches according to the feature maps output by the channel attention module, and the iron flow detection information comprises an iron flow target frame and an iron flow mask obtained through identification.
2. The method of claim 1, wherein the method further comprises:
obtaining tank truck detection information through the target identification detection network;
setting a safe iron receiving area according to the detection information of the tank opening vehicle;
and obtaining a safety detection result of the iron receiving process by confirming whether the iron flow target box is positioned in the safe iron receiving area.
3. The method of claim 2, wherein obtaining a ferrous process safety check result by confirming whether the ferrous flow target box is located within the safe ferrous zone comprises:
if the iron flow target frame is confirmed to be located in the safe iron receiving area, judging that the iron receiving process is safe;
and if at least part of the iron flow target frame is confirmed to be positioned outside the safe iron receiving area, judging that the iron flow has overflow.
4. The method of claim 1, wherein the method further comprises:
acquiring multi-channel taphole infrared video stream data;
extracting a plurality of frames of infrared images from the infrared video stream data;
and obtaining a measurement result of the iron flow temperature by fusing the multi-frame infrared image and the iron flow detection information.
5. The method of claim 4, wherein the obtaining a measured iron temperature measurement by fusing the plurality of frames of infrared images and the iron flow detection information comprises:
acquiring image scale factors of an infrared image and a visible light image, wherein the visible light image is obtained by analyzing the multi-channel taphole video stream data;
acquiring the pixel offset of any frame of infrared image and the visible light image of the corresponding frame;
and superposing the iron flow detection information to the infrared image according to the image scale factor and the pixel offset to finish temperature measurement.
6. The method of claim 5, wherein acquiring image scale factors for the infrared image and the visible image comprises: and calculating the pixel difference between the infrared image and the visible light image between every two circle centers of the calibration plate by adopting a calibration method to obtain the scaling of the image, wherein the scaling of the image is used for representing the image scale factor.
7. The method of claim 6, wherein the obtaining the pixel offset of the infrared image of any frame from the visible light image of the corresponding frame comprises: and calculating the pixel difference of corresponding points according to the pixel coordinate positions of the circular coordinate positions in the calibration plate in the infrared image and the visible light image to obtain the pixel offset.
8. The method according to any one of claims 1-7, wherein said decoding the multiple channels of taphole video stream data to obtain multiple frames of taphole images comprises: and decoding the multi-channel taphole video stream data through a GPU decoding frame to obtain the multi-frame taphole image.
9. The method of claim 1, wherein the target recognition detection network is obtained by training a recognition detection network, wherein during the training process, the mean and variance of the same batch of data are solved, and then normalization is performed, and the mean and variance used during the inference stage are the mean and variance expected values of the whole training sample, which are used as the mean and variance of the BN when we perform inference.
10. The method of claim 1, wherein the channel attention module determines the importance of each pixel based on a scale factor in batch normalization.
11. An apparatus for detecting a state of being ironed, the apparatus comprising:
the multi-channel taphole video stream data acquisition module is configured to acquire multi-channel taphole video stream data;
the decoding module is configured to decode the multi-channel taphole video stream data to obtain a plurality of frame taphole images, wherein each taphole corresponds to at least one frame of taphole image;
the iron flow detection information acquisition module is configured to input the multiple frames of iron notch images into a target identification detection network, and obtain iron flow detection information of at least part of iron notches through the target identification detection network, wherein the target identification detection network comprises a multi-scale feature extraction module, a channel attention module and a prediction regression module which are sequentially connected, the multi-scale feature extraction module is configured to acquire feature maps of the same iron notch in different scales based on the input iron notch images, the channel attention module is used for enhancing the channel attention of the feature maps, the prediction regression module is configured to acquire the iron flow detection information of at least part of iron notches according to the feature maps output by the channel attention module, and the iron flow detection information comprises an identified iron flow target frame and an iron flow mask.
12. A computer program product, characterized in that the computer program product comprises a computer program, wherein the computer program when executed by a processor is adapted to perform the method of any of claims 1-10.
13. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, is adapted to carry out the method of any one of claims 1 to 10.
14. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program is adapted to implement the method of any of claims 1-10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111531771.XA CN114187280A (en) | 2021-12-14 | 2021-12-14 | Method and device for detecting iron receiving state |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111531771.XA CN114187280A (en) | 2021-12-14 | 2021-12-14 | Method and device for detecting iron receiving state |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114187280A true CN114187280A (en) | 2022-03-15 |
Family
ID=80605077
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111531771.XA Pending CN114187280A (en) | 2021-12-14 | 2021-12-14 | Method and device for detecting iron receiving state |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114187280A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929578A (en) * | 2019-10-25 | 2020-03-27 | 南京航空航天大学 | Anti-blocking pedestrian detection method based on attention mechanism |
| CN112257590A (en) * | 2020-10-22 | 2021-01-22 | 中冶南方工程技术有限公司 | Automatic detection method and system for working state of blast furnace taphole and storage medium |
| US20210357763A1 (en) * | 2020-05-18 | 2021-11-18 | StradVision, Inc. | Method and device for performing behavior prediction by using explainable self-focused attention |
-
2021
- 2021-12-14 CN CN202111531771.XA patent/CN114187280A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929578A (en) * | 2019-10-25 | 2020-03-27 | 南京航空航天大学 | Anti-blocking pedestrian detection method based on attention mechanism |
| US20210357763A1 (en) * | 2020-05-18 | 2021-11-18 | StradVision, Inc. | Method and device for performing behavior prediction by using explainable self-focused attention |
| CN112257590A (en) * | 2020-10-22 | 2021-01-22 | 中冶南方工程技术有限公司 | Automatic detection method and system for working state of blast furnace taphole and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9047518B2 (en) | Method for the detection and tracking of lane markings | |
| CN116420159A (en) | Defect detection method, device and system | |
| Ragothaman et al. | Unsupervised segmentation of cervical cell images using Gaussian mixture model | |
| CN102582664A (en) | Congestion estimating apparatus | |
| CN100588959C (en) | Small vehicle holding articles automatic detection method based on radiation image-changing testing | |
| US20130301876A1 (en) | Video analysis | |
| CN114898249B (en) | Method, system and storage medium for confirming number of articles in shopping cart | |
| CN115862259B (en) | Fire alarm early warning system based on temperature monitoring | |
| JP5262705B2 (en) | Motion estimation apparatus and program | |
| CN108830131B (en) | Deep learning-based traffic target detection and ranging method | |
| CN112734750B (en) | Tank cover visual detection and positioning method and system, medium and device | |
| CN111079621A (en) | Method and device for detecting object, electronic equipment and storage medium | |
| JP2018190082A (en) | Vehicle model discrimination device, vehicle model discrimination method, and vehicle model discrimination system | |
| KR20210138367A (en) | Device and Method for estimating of the probability of traffic accidents by road shape analysis | |
| Chaddad et al. | Carcinoma cell identification via optical microscopy and shape feature analysis | |
| CN110705553A (en) | Scratch detection method suitable for vehicle distant view image | |
| Sun et al. | Unsupervised surface defect detection of aluminum sheets with combined bright-field and dark-field illumination | |
| CN102013007A (en) | Apparatus and method for detecting face | |
| CN114187280A (en) | Method and device for detecting iron receiving state | |
| CN116110006B (en) | Scenic spot tourist abnormal behavior identification method for intelligent tourism system | |
| CN116503912A (en) | Security check early warning method based on electronic graph bag | |
| CN110796099A (en) | Vehicle overrun detection method and device | |
| CN115908275A (en) | Hot ring rolling deformation geometric state online measurement method based on deep learning | |
| CN114882552A (en) | Method for checking wearing state of person mask in operation vehicle based on deep learning | |
| Wibowo et al. | Implementation of Background Subtraction for Counting Vehicle Using Mixture of Gaussians with ROI Optimization |
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: 20220315 |
|
| RJ01 | Rejection of invention patent application after publication |