CN117455891A - Appearance detection method and system for air conditioner controller - Google Patents
Appearance detection method and system for air conditioner controller Download PDFInfo
- Publication number
- CN117455891A CN117455891A CN202311547036.7A CN202311547036A CN117455891A CN 117455891 A CN117455891 A CN 117455891A CN 202311547036 A CN202311547036 A CN 202311547036A CN 117455891 A CN117455891 A CN 117455891A
- Authority
- CN
- China
- Prior art keywords
- air conditioner
- conditioner controller
- picture
- appearance
- detection
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 124
- 230000007547 defect Effects 0.000 claims abstract description 23
- 238000000034 method Methods 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 11
- 238000011176 pooling Methods 0.000 claims description 10
- 238000012937 correction Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 239000011521 glass Substances 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
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
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/06—Recognition of objects for industrial automation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an appearance detection method and system of an air conditioner controller, wherein pictures are collected, preprocessed, segmented into appearance pictures of the air conditioner controller, corrected and compared with qualified template pictures through pixels, unqualified products are picked out after the first detection, and key parts of the air conditioner controller are mainly detected after the first detection; and secondly, a trained fast-RCNN network model is used, appearance pictures of the air conditioner controller with defects are detected for the second time, the glass display part of the air conditioner controller is detected mainly through the second detection, the accuracy and the detection efficiency of the detection are improved effectively through the second detection, and the problems of low accuracy and low detection efficiency in the detection of the appearance of the air conditioner controller in the prior art are solved.
Description
Technical Field
The invention relates to the field of machine vision, in particular to an appearance detection method and system for an air conditioner controller.
Background
In the generation process of the air conditioner controller, in order to ensure the quality of the air conditioner controller, the appearance of the air conditioner controller is usually required to be detected, and unqualified products are avoided, wherein the appearance of the air conditioner controller comprises scratch, chromatic aberration, bubbles and the like. At present, the appearance detection of the air conditioner controller usually needs to be manually detected, the detection is affected by subjective factors by adopting the manual detection, and the cost of the manual detection is higher; however, the method of simply adopting deep learning requires a large amount of training data, which is limited in practical application.
Disclosure of Invention
The invention provides an appearance detection method and system for an air conditioner controller, and aims to solve the problems of low accuracy and low detection efficiency when the appearance of the air conditioner controller is detected in the prior art.
In order to achieve the above purpose, the invention adopts the following technical scheme:
the invention provides an appearance detection method of an air conditioner controller, which comprises the following steps:
dividing the acquired picture to obtain an appearance picture of the air conditioner controller, and correcting the position of the appearance picture of the air conditioner controller;
performing template matching on the appearance picture of the air conditioner controller after the position correction and the picture of the standard air conditioner controller to obtain a first detection result;
when the first detection result shows that the appearance picture of the air conditioner controller is qualified, inputting the appearance picture of the air conditioner controller into a trained fast-RCNN network model, and carrying out second detection on the appearance picture of the air conditioner controller to obtain a second detection result;
and when the second detection result shows that the appearance picture of the air conditioner controller is qualified, the current air conditioner controller is considered to be a qualified product.
According to the invention, the pictures are collected, the pictures are divided, the appearance pictures of the air conditioner controller are divided, the position of the appearance pictures of the air conditioner controller is corrected, template matching is carried out on the pictures with the standard air conditioner controller, unqualified products are picked out after the first detection, and the key part of the air conditioner controller is mainly detected after the first detection; and secondly, a trained fast-RCNN network model is used for carrying out secondary detection on the appearance picture of the air conditioner controller which is qualified in the primary detection, the secondary detection mainly detects the glass display part of the air conditioner controller, the accuracy and the detection efficiency of the detection are effectively improved through the secondary detection, and the problems of low accuracy and low detection efficiency in the detection of the appearance of the air conditioner controller in the prior art are solved.
Optionally, the training process of the fast-RCNN network model includes:
collecting appearance pictures of the air conditioner controller with defects in appearance, classifying and marking the appearance pictures of the air conditioner controller according to the types of the defects, and constructing an appearance picture data set of the air conditioner controller;
dividing the appearance picture training data set of the air conditioner controller into a training set and a testing set;
constructing a VGG-16-based Faster-RCNN detection network model, and training the Faster-RCNN detection network model by using the divided data set to obtain a trained Faster-RCNN detection network model.
Optionally, the method for detecting the appearance of the air conditioner controller further comprises the steps of counting qualified product data and unqualified product data, and outputting the counted data as a result.
Optionally, the fast-RCNN detection network model includes:
backbone network: extracting input picture characteristics;
RPN network: receiving the input picture characteristics and constructing a detection target candidate region;
region of interest pooling layer: mapping the detection target candidate region to a characteristic region, and pooling the characteristic region into a uniform scale;
classification and regression layer: and classifying each target category of the characteristic region, and correcting the target frame by utilizing the bounding box regression to obtain the position offset.
Optionally, the method for detecting the appearance of the air conditioner controller provided by the invention further comprises the steps of preprocessing the acquired picture:
graying the picture;
carrying out picture enhancement on the grey picture;
performing binarization processing on the enhanced picture;
and carrying out noise elimination treatment on the binarized picture through Gaussian filtering.
Optionally, the collected pictures are a front picture and a back picture of the air conditioner controller.
The invention also provides an appearance detection system of the air conditioner controller, which comprises:
and a picture processing module: dividing the acquired picture to obtain an appearance picture of the air conditioner controller, and correcting the position of the appearance picture of the air conditioner controller;
the first detection module: performing template matching on the appearance picture of the air conditioner controller after the position correction and the picture of the standard air conditioner controller to obtain a first detection result;
the second detection module: when the first detection result shows that the appearance picture of the air conditioner controller is qualified, inputting the appearance picture of the air conditioner controller into a trained fast-RCNN network model, and carrying out second detection on the appearance picture of the air conditioner controller to obtain a second detection result;
and a judging module: and when the second detection result shows that the appearance picture of the air conditioner controller is qualified, the current air conditioner controller is considered to be a qualified product.
Compared with the prior art, the invention has the beneficial effects that:
(1) According to the invention, the pictures are collected, preprocessed, segmented into the appearance pictures of the air conditioner controller, corrected and compared with the qualified template pictures through pixels, and unqualified products are picked out after the first detection, and the key part of the air conditioner controller is mainly detected after the first detection; and secondly, a trained fast-RCNN network model is used, appearance pictures of the air conditioner controller with defects are detected for the second time, the glass display part of the air conditioner controller is detected mainly through the second detection, the accuracy and the detection efficiency of the detection are improved effectively through the second detection, and the problems of low accuracy and low detection efficiency in the detection of the appearance of the air conditioner controller in the prior art are solved.
(2) The method provided by the invention can meet different defects for detecting the appearance of the air conditioner controller by detecting the appearance picture of the air conditioner controller twice, and reduces the detection cost.
(3) According to the invention, by combining the machine vision and the deep learning model, the detection by manpower is avoided, the problems of high misjudgment rate and high labor cost caused by manual detection are solved, the manpower is saved, and the detection accuracy is improved.
Drawings
Fig. 1 is a schematic diagram of a detection flow of an appearance detection method of an air conditioner controller according to an embodiment of the present invention;
fig. 2 is a flow chart of an appearance detection method of an air conditioner controller according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a training process of a Faster-RCNN network model of an appearance detection method of an air conditioner controller according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a Faster-RCNN network model of an appearance detection method of an air conditioner controller according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an appearance detection system of an air conditioner controller according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made more apparent and fully hereinafter with reference to the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by one of ordinary skill in the art without undue burden on the person of ordinary skill in the art based on embodiments of the present invention, are within the scope of the present invention.
Example 1
The invention provides an appearance detection method and system of an air conditioner controller, which mainly detects the front and the back of the air conditioner controller, wherein the front of the air conditioner controller comprises a glass display and text keys, and the two parts are difficult to detect simultaneously by a general detection algorithm; and secondly, a trained fast-RCNN network model is used, a picture with a defect is detected for the second time, a glass display part of the air conditioner controller is detected mainly for the second time, the accuracy and the detection efficiency of the detection are effectively improved through the twice detection, and meanwhile, the appearance detection of the air conditioner controller designed by the invention comprises high-performance computing equipment and professional detection equipment.
As shown in fig. 1 and 2, according to an aspect of the present embodiment, an appearance detection method of an air conditioner controller is provided, and the method and system provided by the present invention are constructed based on template matching and a fast-RCNN algorithm, and specifically implemented according to the following steps:
s1, firstly, obtaining pictures through an industrial camera, wherein the pictures comprise a front side part and a back side part, respectively photographing the front side and the back side of an air conditioner controller by fixing the two industrial cameras above a conveyor belt, putting the air conditioner controller to be detected into the conveyor belt, obtaining the pictures of the front side and the back side of the air conditioner controller to be detected, and storing the pictures in a database.
S2, because the obtained picture is influenced by objective factors, preprocessing operation is needed to be carried out on the collected front and back images of the appearance of the air conditioner controller, and three-component brightness in the color image is averaged to obtain a gray value to carry out image graying on the picture, as shown in a formula 1:
Gray(i,j)＝(R(i,j)+G(i+j)+B(i,j))/3 (1)
the image of the air conditioner controller is a low-gray image, the image is corrected through gamma change, the contrast is enhanced through enhancing the details of low gray, and a change formula is that each pixel on the image is subjected to product operation, as shown in a formula 2:
S＝C*a^n(2)
wherein c and n are constants
After the gamma change is enhanced, a self-adaptive threshold method is adopted to carry out binarization processing, a local area is selected, the gray level is solved in the local area, the median is selected as a threshold value, all pixel points in the local area are compared with the threshold value, if the pixel point is larger than the threshold value, the pixel point is white, and if the pixel point is smaller than the threshold value, the pixel point is black. The self-adaptive threshold method determines a local self-adaptive threshold based on the pixel gray value in the pixel field, and the threshold is automatically selected and adjusted according to the local brightness, contrast and texture information of the image, so that the details of the appearance characters of the air conditioner controller can be improved; and finally, eliminating noise of the whole image through a Gaussian filter.
S3, dividing the preprocessed air conditioner controller picture, dividing the picture by adopting a global threshold value, setting a fixed threshold value, classifying the picture as a foreground if the pixel value of the picture is larger than the threshold value, classifying the picture as a background if the pixel value of the picture is smaller than the threshold value, and setting the foreground pixel as white and the background pixel as black; and obtaining a segmented image of the air conditioner controller, and removing the saw-tooth or burr-shaped effect by smoothing the edge through Gaussian filtering, so as to ensure the detection accuracy.
S4, after the step of complaining is finished, the segmented air conditioner controller can be subjected to image correction through perspective change, the images are corrected to the same position, subsequent feature detection is facilitated, and meanwhile the detection accuracy can be improved. The change formula of perspective transformation is as follows:
[ x y z ] is an image after correction
[ uvw ] is the original image pixel coordinates
T1 is the linear change matrix of the image
T2 is a nonlinear variation matrix
T3 is an image translation matrix
And S5, after the image is corrected to the unified position, performing template matching. And randomly taking a plurality of qualified air conditioner controllers, carrying out unified processing through the steps, averaging the obtained pixels to obtain standard pixels, and comparing the pixels of the image with the pixels of the appearance of the standard air conditioner controllers one by one. Since small errors in image detection may be due to noise, a suitable threshold should be set and when the error rate of detection is greater than the threshold, a detection defect is identified. The detection is mainly used for detecting whether defects appear in the appearance size area, the color, the radian, the keys and the key words of the air conditioner controller.
And S6, when the appearance defect is not detected by template matching, further detecting the appearance of the air conditioner controller by using a trained Faster-RCNN network model, wherein the training process is shown in a figure 3, and the model structure of the Faster-RCNN is shown in a figure 4.
The Faster-RCNN air conditioner controller appearance detection model for training the air conditioner controller appearance detection comprises the following steps:
s6-1, acquiring a front picture and a back picture of an air conditioner controller with a defective appearance, storing all pictures in jpg format, naming the front picture as an image 1_sequence number, and naming the back picture as an image 0_sequence number;
s6-2, making a label, classifying defects of the obtained picture, marking the type of the picture by using manual marking software Labelimg, and generating an xml file with the same name as the picture file, wherein the xml file comprises 4 position attributes and 1 type attribute;
s6-3, data expansion, namely, performing data enhancement with xml files on a data set which is marked recently because deep learning requires a large number of air conditioner controller appearance defect pictures to perform feature learning, and expanding the data by using rotation of 45 degrees, 90 degrees and 180 degrees;
s6-4, constructing a data set, and setting the expanded data to 8:2 is divided into a training set and a testing set;
s6-5, training a model, namely building a fast-RCNN network structure by using VGG16 as a main network, setting the learning rate to be 0.001, setting the batch_size to be 16, setting the maximum iteration number to be 10000, setting dynamic descent to enable the model to be better converged, gradually descending a loss value in the training process, and finally achieving the convergence effect;
s6-6, performing appearance detection of the air conditioner controller, performing secondary detection when the defects are not detected by template matching, detecting images through a trained fast-RCNN network model, identifying pictures with suspected defects of the appearance of the air conditioner controller by detection, giving a prediction frame and confidence level to the pictures with defect characteristics by the model, and recognizing that the appearance of the air conditioner controller has the defects when the confidence level is greater than 0.95.
The training process of the S6-5, faster-RCNN network model is shown in FIG. 3. The specific process of training the model is as follows: and extracting features of the picture by using CNN, acquiring suggestions and region scores by using an RPN sliding window mechanism, outputting the first N high-score interval suggestions by using a non-maximum suppression algorithm, inputting the region suggestions into an ROI pooling layer, obtaining region suggestion features by using the pooling layer, and finally classifying the targets by using a full-connection layer.
Further shown as S6-5 includes the following processes:
building a fast-RCNN network model to train the training set, setting the super-parameters of training, setting the learning rate to be 0.001, setting the maximum iteration number to be 10000, setting the feature extraction model to be VGG16, and setting the batch processing size to be 16.
And (3) inputting the picture into a neural network, extracting the characteristics of the picture by using VGG16 of a preserving convolution layer, a pooling layer and a ReLU activation function layer, and inputting the characteristics into a final full convolution network and a full connection layer.
And outputting a series of rectangular region candidate frames for each input image through a full convolution network RPN network, and simultaneously containing the confidence coefficient of a corresponding target, and obtaining a feature map with a fixed size according to the region candidate frames generated by the region generating network and the feature map obtained before.
The method is mainly used for distinguishing defect types and backgrounds, outputting probability vectors, obtaining the position offset of each suggested frame by utilizing bounding box regression, calculating a loss function, and updating the whole network parameters to obtain a training model.
The RPN module in the step (3) is mainly used for generating candidate areas, firstly, traversing feature mapping by using a sliding window with the size of 3 multiplied by 3 to generate a plurality of anchor frames (anchors), judging the class of the anchor frames (anchors) through a function after cutting and filtering the anchor frames, correcting the anchors by using frame regression to obtain accurate candidate areas, collecting the candidate areas by using Region of interest pooling (ROI), calculating a feature map candidate area, and sending the feature map candidate area to a full-connection layer to judge the target class.
And detecting through a secondary fast-RCNN network model, and mainly detecting the fine defects of the air conditioner controller, such as stains, scratches, whether a display screen is damaged or not and the like. And if no defect is detected through the two detection, the appearance of the air conditioner controller is qualified.
In summary, the invention performs the appearance defect screening on the air conditioner controller for the first time through template matching, and performs further detail detection through the fast-RCNN network model, and the product is qualified after the two detection. And feeding the unqualified products back to the user by the system, and counting the unqualified rate.
Example 2
Based on the same inventive concept as embodiment 1, this embodiment introduces an air conditioner controller appearance detection system, including:
and a picture processing module: dividing the acquired picture to obtain an appearance picture of the air conditioner controller, and correcting the position of the appearance picture of the air conditioner controller;
the first detection module: performing template matching on the appearance picture of the air conditioner controller after the position correction and the picture of the standard air conditioner controller to obtain a first detection result;
the second detection module: when the first detection result shows that the appearance picture of the air conditioner controller is qualified, inputting the appearance picture of the air conditioner controller into a trained fast-RCNN network model, and carrying out second detection on the appearance picture of the air conditioner controller to obtain a second detection result;
and a judging module: and when the second detection result shows that the appearance picture of the air conditioner controller is qualified, the current air conditioner controller is considered to be a qualified product.
The specific function implementation of each module is described in the method of reference embodiment 1, and is not repeated, and specifically noted is that:
the image processing module further comprises preprocessing the acquired image, wherein the preprocessing of the image is completed through the following steps: graying the acquired picture; carrying out picture enhancement on the grey picture; performing binarization processing on the enhanced picture; and carrying out noise elimination treatment on the binarized picture through Gaussian filtering.
The training process of the trained fast-RCNN network model in the second detection module is as follows: collecting appearance pictures of the air conditioner controller with defects in appearance, classifying and marking the appearance pictures of the air conditioner controller according to the types of the defects, and constructing an appearance picture data set of the air conditioner controller; dividing an appearance picture training data set of the air conditioner controller into a training set and a testing set; constructing a VGG-16-based Faster-RCNN detection network model, and training the Faster-RCNN detection network model by using the divided data set to obtain a trained Faster-RCNN detection network model.
And the Faster-RCNN network model comprises: backbone network: extracting input picture characteristics; RPN network: receiving the input picture characteristics and constructing a detection target candidate region; region of interest pooling layer: mapping the detection target candidate region to a characteristic region, and pooling the characteristic region into a uniform scale; classification and regression layer: and classifying each target category of the characteristic region, and correcting the target frame by utilizing the bounding box regression to obtain the position offset.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (7)
1. An appearance detection method of an air conditioner controller is characterized by comprising the following steps:
dividing the acquired picture to obtain an appearance picture of the air conditioner controller, and correcting the position of the appearance picture of the air conditioner controller;
performing template matching on the appearance picture of the air conditioner controller after the position correction and the picture of the standard air conditioner controller to obtain a first detection result;
when the first detection result shows that the appearance picture of the air conditioner controller is qualified, inputting the appearance picture of the air conditioner controller into a trained fast-RCNN network model, and carrying out second detection on the appearance picture of the air conditioner controller to obtain a second detection result;
and when the second detection result shows that the appearance picture of the air conditioner controller is qualified, the current air conditioner controller is considered to be a qualified product.
2. The method for detecting the appearance of an air conditioner controller according to claim 1, wherein the training process of the fast-RCNN network model comprises:
collecting appearance pictures of the air conditioner controller with defects in appearance, classifying and marking the appearance pictures of the air conditioner controller according to the types of the defects, and constructing an appearance picture data set of the air conditioner controller;
dividing the appearance picture training data set of the air conditioner controller into a training set and a testing set;
constructing a VGG-16-based Faster-RCNN detection network model, and training the Faster-RCNN detection network model by using the divided data set to obtain a trained Faster-RCNN detection network model.
3. The air conditioner controller appearance detection method of claim 2, wherein the fast-RCNN detection network model comprises:
backbone network: extracting input picture characteristics;
RPN network: receiving the input picture characteristics and constructing a detection target candidate region;
region of interest pooling layer: mapping the detection target candidate region to a characteristic region, and pooling the characteristic region into a uniform scale;
classification and regression layer: and classifying each target category of the characteristic region, and correcting the target frame by utilizing the bounding box regression to obtain the position offset.
4. The method for detecting the appearance of an air conditioner controller according to claim 1, further comprising counting product pass data and product fail data, and outputting the counted data as a result.
5. The method for detecting the appearance of an air conditioner controller according to claim 1, further comprising preprocessing the acquired picture:
graying the picture;
carrying out picture enhancement on the grey picture;
performing binarization processing on the enhanced picture;
and carrying out noise elimination treatment on the binarized picture through Gaussian filtering.
6. The method for detecting the appearance of an air conditioner controller according to claim 1, wherein the acquired pictures are a front picture and a back picture of the air conditioner controller.
7. An appearance detection system of an air conditioner controller, comprising:
and a picture processing module: dividing the acquired picture to obtain an appearance picture of the air conditioner controller, and correcting the position of the appearance picture of the air conditioner controller;
the first detection module: performing template matching on the appearance picture of the air conditioner controller after the position correction and the picture of the standard air conditioner controller to obtain a first detection result;
the second detection module: when the first detection result shows that the appearance picture of the air conditioner controller is qualified, inputting the appearance picture of the air conditioner controller into a trained fast-RCNN network model, and carrying out second detection on the appearance picture of the air conditioner controller to obtain a second detection result;
and a judging module: and when the second detection result shows that the appearance picture of the air conditioner controller is qualified, the current air conditioner controller is considered to be a qualified product.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311547036.7A CN117455891A (en) | 2023-11-20 | 2023-11-20 | Appearance detection method and system for air conditioner controller |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311547036.7A CN117455891A (en) | 2023-11-20 | 2023-11-20 | Appearance detection method and system for air conditioner controller |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117455891A true CN117455891A (en) | 2024-01-26 |
Family
ID=89581750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311547036.7A Pending CN117455891A (en) | 2023-11-20 | 2023-11-20 | Appearance detection method and system for air conditioner controller |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117455891A (en) |
-
2023
- 2023-11-20 CN CN202311547036.7A patent/CN117455891A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107543828B (en) | Workpiece surface defect detection method and system | |
| CN110992329B (en) | Product surface defect detection method, electronic equipment and readable storage medium | |
| CN111833306B (en) | Defect detection method and model training method for defect detection | |
| CN106875381B (en) | Mobile phone shell defect detection method based on deep learning | |
| CN110390677B (en) | Defect positioning method and system based on sliding self-matching | |
| CN111915704A (en) | Apple hierarchical identification method based on deep learning | |
| CN106846316A (en) | A kind of GIS inside typical defect automatic distinguishing method for image | |
| CN113706490B (en) | Wafer defect detection method | |
| CN114926407A (en) | Steel surface defect detection system based on deep learning | |
| CN112750113B (en) | Glass bottle defect detection method and device based on deep learning and linear detection | |
| CN115829995A (en) | Cloth flaw detection method and system based on pixel-level multi-scale feature fusion | |
| CN114004858A (en) | Method and device for identifying aviation cable surface code based on machine vision | |
| CN114298985B (en) | Defect detection method, device, equipment and storage medium | |
| CN114612418A (en) | Method, device and system for detecting surface defects of mouse shell and electronic equipment | |
| CN111178405A (en) | Similar object identification method fusing multiple neural networks | |
| CN117011300B (en) | Micro defect detection method combining instance segmentation and secondary classification | |
| CN116797602A (en) | Surface defect identification method and device for industrial product detection | |
| CN112488986A (en) | Cloth surface flaw identification method, device and system based on Yolo convolutional neural network | |
| CN114548250B (en) | Mobile phone appearance detection method and device based on data analysis | |
| CN110889418A (en) | Gas contour identification method | |
| CN117455891A (en) | Appearance detection method and system for air conditioner controller | |
| CN112686851B (en) | Image detection method, device and storage medium | |
| CN112150434A (en) | Tire defect detection method, device, equipment and storage medium | |
| CN113569737B (en) | Notebook screen defect detection method and medium based on autonomous learning network model | |
| CN113066075B (en) | Multi-image fusion denim flaw detection method and device |
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 |