CN112179846A - Prefabricated convex window defect detection system based on improved Faster R-CNN - Google Patents
Prefabricated convex window defect detection system based on improved Faster R-CNN Download PDFInfo
- Publication number
- CN112179846A CN112179846A CN202011244564.1A CN202011244564A CN112179846A CN 112179846 A CN112179846 A CN 112179846A CN 202011244564 A CN202011244564 A CN 202011244564A CN 112179846 A CN112179846 A CN 112179846A
- Authority
- CN
- China
- Prior art keywords
- fixed cover
- cnn
- defect detection
- prefabricated
- fixedly connected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/01—Arrangements or apparatus for facilitating the optical investigation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Biochemistry (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a prefabricated bay window defect detection system based on improved fast R-CNN, which comprises an image collection device and a background PC, wherein the image collection device comprises a screw rod, the left end and the right end of the surface of the screw rod are both in threaded connection with a sliding nut, the left side and the right side of the sliding nut are both fixedly connected with an extension rod, the surface of the extension rod is respectively sleeved with a first fixing sleeve, a second fixing sleeve and a third fixing sleeve from the direction close to the screw rod to the direction far away from the screw rod, and springs are respectively and fixedly connected between the first fixing sleeve and the second fixing sleeve and between the second fixing sleeve and the third fixing sleeve and positioned on the surface of the extension rod. This prefabricated bay window defect detecting system based on improve Faster R-CNN through being provided with the lead screw, cooperation extension rod, first fixed cover, the fixed cover of second, the fixed cover of third, spring, industry camera, can realize carrying out image acquisition to the prefabricated bay window of different specifications to every industrial camera position of collecting in the group all can self-adaptation regulation.
Description
Technical Field
The invention relates to the technical field, in particular to a prefabricated convex window defect detection system based on improved Faster R-CNN.
Background
With the development of computer technology and image processing technology, machine vision inspection technology is widely applied to industrial defect inspection. For example: celik et al combine the image algorithm based on bithreshold binarization, wavelet transform and morphological operations with a feedforward neural network to realize defect detection; affhar et al use statistical methods to detect defect edges and then identify defects using a multi-strategy support vector machine as a classifier; guo Hui et al propose a method for achieving defect detection based on a gray level co-occurrence matrix and a hierarchical clustering algorithm.
In recent years, the application of deep learning technology in the industry as an important branch of machine learning is becoming mature. YJ Cha et al uses a convolutional neural network to identify concrete crack defects; FC Chen et al propose a deep learning framework based on naive Bayes and convolutional neural network data fusion, used for detecting crack defects; the Yuyongwei et al extracts the essential features of the suspicious defect region using a convolutional neural network, and then identifies the defect using a radial basis network.
The existing quality is the life of a company, the product quality and the production efficiency become the most important of the development of the company, so the quality detection of the prefabricated convex window is vital to the quality guarantee of the prefabricated convex window, the surface of the product possibly has defects in the daily production process of the prefabricated convex window, the traditional surface detection method mainly adopts manual visual inspection, the long-time visual inspection is easy to generate fatigue, the false inspection rate is increased, the working efficiency is reduced, the problems of large workload, low detection efficiency, high false inspection rate and the like exist in the manual visual detection method, the traditional machine vision detection method usually adopts an industrial camera to acquire images, the acquired images are preprocessed by using a digital image technology, then a classifier is set to classify and detect the defective images, the detection model is single, and certain key parameters need to be manually manufactured or manually selected, the detection method is greatly influenced by human subjectivity, so that the generalization capability of the detection method is not high, and uncontrollable factors such as multiple defects of the prefabricated convex window, complex production environment and the like cannot be responded.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a prefabricated convex window defect detection system based on improved Faster R-CNN, which solves the problems that the manual visual detection method is large in workload, low in detection efficiency, high in false detection rate and the like, the traditional machine vision detection method usually adopts an industrial camera to acquire images, the detection model is single, certain key parameters need to be manually manufactured or manually selected, the influence of artificial subjectivity is large, the generalization capability of the detection method is not high, the uncontrollable factors such as multiple prefabricated convex window defects, complex production environment and the like cannot be met.
In order to achieve the purpose, the invention is realized by the following technical scheme: the utility model provides a prefabricated bay window defect detecting system based on improve fast R-CNN, includes image collection device and backstage PC, image collection device includes the lead screw, the equal threaded connection in both ends has sliding nut about the lead screw surface, the equal fixedly connected with extension rod in the left and right sides of sliding nut, the surface of extension rod is equipped with the fixed cover of first fixed cover, second and the fixed cover of third from being close to the lead screw to the direction of keeping away from the lead screw respectively, the fixed cover of first fixed cover and second with between the fixed cover of second and the fixed cover of third between and be located the equal fixedly connected with spring in surface of extension rod, the fixed cover of first fixed cover, second and the fixed equal fixedly connected with industrial camera in surface of third.
Preferably, both ends are provided with the screw of equidirectional not, the middle fixedly connected with gag lever post on lead screw surface, the left and right sides of gag lever post rotates and is connected with the connecting rod, one of them the bottom of connecting rod rotates with the top of the fixed cover of third to be connected, another the bottom of connecting rod rotates with the top of first fixed cover to be connected.
Preferably, one the extension rod, a first fixed cover, a second fixed cover, a third fixed cover, two springs, three industrial cameras, a gag lever post and two connecting rods form a set of collection group, image collection device is provided with four sets of collection groups, and two sets of adjacent collection groups are mutually perpendicular.
Preferably, a base plate is arranged on the front side of the screw rod, a servo motor is fixedly connected to the rear side of the base plate, and the output end of the servo motor is fixedly connected with the screw rod.
Preferably, the front side of the substrate is fixedly connected with a support arm, the bottom end of the support arm is fixedly connected with a base, and the bottom of the base is fixedly connected with a pulley.
Preferably, the background PC is fixed on top of the base, and the background PC comprises an analysis system.
Preferably, the analysis system comprises image preprocessing, improved Faster R-CNN, prefabricated convex window defect detection and prefabricated convex window defect judgment.
Preferably, the output end of the image preprocessing is coupled with the input end of the improved Faster R-CNN, the output end of the improved Faster R-CNN is coupled with the input end of the prefabricated convex window defect detection, and the output end of the prefabricated convex window defect detection is coupled with the input end of the prefabricated convex window defect judgment.
Advantageous effects
The invention provides a prefabricated convex window defect detection system based on improved Faster R-CNN. Compared with the prior art, the method has the following beneficial effects:
(1) the prefabricated bay window defect detection system based on the improved Faster R-CNN comprises an image collection device and a background PC, wherein the image collection device comprises a screw rod, the left end and the right end of the surface of the screw rod are in threaded connection with sliding nuts, the left side and the right side of each sliding nut are fixedly connected with an extension rod, the surface of each extension rod is respectively sleeved with a first fixing sleeve, a second fixing sleeve and a third fixing sleeve from the direction close to the screw rod to the direction far away from the screw rod, springs are respectively and fixedly connected between the first fixing sleeve and the second fixing sleeve and between the second fixing sleeve and the third fixing sleeve and positioned on the surface of each extension rod, the first fixing sleeve, the second fixing sleeve and the third fixing surface are respectively and fixedly connected with an industrial camera, the left end and the right end of the screw rod are provided with threads in different directions, a limiting rod is fixedly connected in the middle of the, the bottom of one of them connecting rod is rotated with the top of the fixed cover of third and is connected, the bottom of another connecting rod is rotated with the top of the fixed cover of first and is connected, an extension rod, a first fixed cover, the fixed cover of a second, the fixed cover of a third, two springs, three industry camera, a gag lever post and two connecting rods form a set of collection group, image collection device is provided with four groups and collects the group, mutually perpendicular between adjacent two sets of collection groups, through being provided with the lead screw, the cooperation extension rod, first fixed cover, the fixed cover of second, the fixed cover of third, a spring, industry camera, can realize carrying out image acquisition to the prefabricated convex window of different specifications, and every group collects the industrial camera position of organizing and all can self-adaptive control.
(2) This prefabricated bay window defect detecting system based on improve Faster R-CNN, through being provided with the base plate in the front side of lead screw, the rear side fixedly connected with servo motor of base plate, servo motor's output and lead screw fixed connection, the front side fixedly connected with support arm of base plate, the bottom fixedly connected with base of support arm, the bottom fixedly connected with pulley of base, through the pulley, mobile device that can be convenient fast, cooperation base plate and servo motor make the regulation more convenient, and work efficiency has been accelerated, and the practicality has been increased.
(3) The prefabricated bay window defect detection system based on the improved Faster R-CNN is fixed on the top of a base through a background PC, the background PC comprises an analysis system, the analysis system comprises image preprocessing, improved Faster R-CNN, prefabricated bay window defect detection and prefabricated bay window defect judgment, the output end of the image preprocessing is coupled with the input end of the improved Faster R-CNN, the output end of the improved Faster R-CNN is coupled with the input end of the prefabricated bay window defect detection, the output end of the prefabricated bay window defect detection is coupled with the input end of the prefabricated bay window defect judgment, the improved Faster R-CNN network is used for realizing the rapid and real-time prefabricated bay window defect detection, the problems of easy fatigue generation and high false detection rate of manual visual inspection and long-time visual inspection are solved, and the product quality inspection efficiency is improved, the method has the advantages that the feature extraction is realized through automatic learning, the problem that the target detection cannot be realized in a complex and changeable production environment through general feature extraction is solved, and compared with an original network, the improved Faster R-CNN network enables small target defects not to be lost easily in the detection process, and the quality detection accuracy of the prefabricated convex window is improved.
Drawings
FIG. 1 is a front view of the present invention;
FIG. 2 is a side view of a substrate according to the present invention;
FIG. 3 is a partial side view of an image capture device of the present invention;
FIG. 4 is an enlarged view of a portion of the invention at A in FIG. 3;
FIG. 5 is a logic flow diagram of the present invention;
FIG. 6 is a schematic diagram of an improved Faster R-CNN network according to the present invention;
FIG. 7 is a schematic diagram of a prefabricated convex window defect prediction network of the fusion feature pyramid in the present invention;
fig. 8 is a schematic diagram of the basic structure of the residual error network in the present invention.
In the figure: 1. an image collection device; 2. a background PC; 101. a screw rod; 102. a sliding nut; 103. an extension rod; 104. a first fixing sleeve; 105. a second fixing sleeve; 106. a third fixing sleeve; 107. a spring; 108. an industrial camera; 109. a limiting rod; 110. a connecting rod; 111. a substrate; 112. a servo motor; 113. a support arm; 114. a base; 115. a pulley; 201. an analysis system; 202. preprocessing an image; 203. modified Faster R-CNN; 204. detecting defects of the prefabricated convex window; 205. and judging the defects of the prefabricated convex window.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1-7, the present invention provides a technical solution: a prefabricated bay window defect detection 204 system based on improved Faster R-CNN comprises an image collection device 1 and a background PC2, wherein the image collection device 1 comprises a screw rod 101, the left and right ends of the surface of the screw rod 101 are both in threaded connection with a sliding nut 102, the left and right sides of the sliding nut 102 are both fixedly connected with an extension rod 103, the surface of the extension rod 103 is respectively sleeved with a first fixing sleeve 104, a second fixing sleeve 105 and a third fixing sleeve 106 from the direction close to the screw rod 101 to the direction far away from the screw rod 101, springs 107 are respectively fixedly connected between the first fixing sleeve 104 and the second fixing sleeve 105 and between the second fixing sleeve 105 and the third fixing sleeve 106 and positioned on the surface of the extension rod 103, the surfaces of the first fixing sleeve 104, the second fixing sleeve 105 and the third fixing sleeve are all fixedly connected with industrial cameras 108, the left and right ends of the screw rod 101 are provided with threads in different directions, a limit rod 109, the left side and the right side of the limiting rod 109 are rotatably connected with connecting rods 110, the bottom end of one connecting rod 110 is rotatably connected with the top of a third fixing sleeve 106, the bottom end of the other connecting rod 110 is rotatably connected with the top of a first fixing sleeve 104, one extending rod 103, one first fixing sleeve 104, one second fixing sleeve 105, one third fixing sleeve 106, two springs 107, three industrial cameras 108, one limiting rod 109 and two connecting rods 110 form a group of collecting groups, the image collecting device 1 is provided with four groups of collecting groups, two adjacent groups of collecting groups are perpendicular to each other, the image collecting device can realize image collection on prefabricated convex windows with different specifications by being provided with a screw 101 and matched with the extending rod 103, the first fixing sleeve 104, the second fixing sleeve 105, the third fixing sleeve 106, the springs 107 and the industrial cameras 108, and the positions of the industrial cameras 108 in each group of collecting groups can be adjusted in a self-, the front side of the screw rod 101 is provided with a substrate 111, the rear side of the substrate 111 is fixedly connected with a servo motor 112, the output end of the servo motor 112 is fixedly connected with the screw rod 101, the front side of the substrate 111 is fixedly connected with a supporting arm 113, the bottom end of the supporting arm 113 is fixedly connected with a base 114, the bottom of the base 114 is fixedly connected with a pulley 115, the device can be moved quickly and conveniently by the pulley 115, the adjustment is more convenient by matching with the substrate 111 and the servo motor 112, the working efficiency is improved, the practicability is increased, a background PC2 is fixed on the top of the base 114, a background PC2 comprises an analysis system 201, the analysis system 201 comprises an image preprocessing 202, an improved Faster R-CNN203, a prefabricated bay window defect detection 204 and a prefabricated bay window defect judgment 205, the output end of the image preprocessing 202 is coupled with the input end of the improved fast R-CNN203, the output end of the improved fast R-CNN203 is coupled with the input end of the prefabricated bay, the output end of the prefabricated convex window defect detection 204 is coupled with the input end of the prefabricated convex window defect judgment 205, the improved Faster R-CNN203 network is used for realizing the rapid and real-time prefabricated convex window defect detection 204, the problems of fatigue and high false detection rate of long-time visual detection due to manual visual inspection are solved, the product quality inspection efficiency is improved, the feature extraction is realized through automatic learning, the problem that the target detection cannot be realized in a complicated and changeable production environment by general feature extraction is solved, and the improved Faster R-CNN203 network is compared with an original network, so that small target defects are not easy to lose in the detection process, and the prefabricated convex window quality inspection accuracy is improved.
Please refer to fig. 8, which shows the following steps:
c1, C2 and C3 are initial feature maps, P3, P2 and P1 are feature maps which are transversely connected and are subjected to 1 × 1 convolution, and then 3 × 3 convolution is used for eliminating aliasing effects caused in the upsampling process, so that possible areas of defects in the images are calibrated.
Wherein x represents an input value, Fx is an output value of the x value after two layers of convolution and one layer of Relu activation function, in ResNet, the input and the output are added together by using skip layer connection, at this time, the unit training target becomes Fx = Hx-x, namely the residual error of the training unit, and the comprehensive output y of the residual error unit is as follows:
wherein, for the Relu activation function, W1 and W2 are the weights of the first layer network and the second layer network, respectively.
Meanwhile, the contents which are not described in detail in the specification belong to the prior art which is known by the person skilled in the art, and the model parameters of each electric appliance are not particularly limited, and conventional equipment can be used.
During the use, promote pulley 115, drive whole device and remove, reach suitable position after, open servo motor 112, then drive lead screw 101 and rotate, drive sliding nut 102 and remove on the surface of lead screw 101, and when sliding nut 102 drove extension rod 103 and remove, connecting rod 110 was dragged to gag lever post 109 to make first fixed cover 104 and third fixed cover 106 be close to second fixed cover 105, open industrial camera 108 after that, collect the image, image data is transmitted to in the analytic system 201.
Step 1: prefabricated bay window image acquisition
The industrial camera 108 is arranged at the front end, long connection is established between the industrial camera and the background PC2 end, and image data collected regularly are transmitted to the background PC2 end through Socket.
Step 2: image pre-processing 202
The image preprocessing 202 maximizes restoration of the original features of the image and minimizes useless information such as noise irrelevant to detection.
2.1 image enhancement
2.1.1 image denoising
The image noise is processed by median filtering, the gray value of a target point is set as the median of all points in a certain area around the target point, the pixel value in the field of the target point is enabled to be closer to the true value, and therefore the independent noise signal is eliminated.
2.1.2 image sharpening Process
After the image is filtered, the definition of the image contour and the edge can be reduced, and the laplacian operator is adopted for sharpening, so that the difference between the gray values of the object and the background in the image is enhanced.
2.2 Deaveraging and normalization Process
And (3) calling a function library to average and normalize the image data, and mapping the data between-1 and 1, so that the data can be rapidly converged and an optimal solution can be found during the training of the neural network model.
And step 3: defect detection
According to the target detection technology and the characteristics of the prefabricated convex window surface defect image, an improved Faster R-CNN203 network is adopted, the FPN is used for fusing multi-scale feature maps to generate feature maps with high-level and low-level semantemes, the RPN provides a series of possible defect candidate areas in the image, the ResNet101 residual error network is used for adding input and output together through 'layer jump connection' to extract image features, and finally the fast R-CNN is used for classifying the feature maps fused with all information to finish target classification and regression tasks.
The feature pyramid network fuses the multi-scale feature maps, reduces the loss of small target defects in training, and improves the defect detection generalization capability. The method mainly comprises three parts, wherein the first part is subjected to up-sampling through a convolutional network, the second part is subjected to down-sampling from the starting of a feature map, the third part is transversely connected, and the feature maps of the first two parts are fused to obtain a multi-scale feature map.
The traditional VGG16 network is used as the basic feature extraction network, gradient dispersion or disappearance is easily caused in the deep network back propagation process, and small target defects are easily lost in the detection process, so the deep residual error network ResNet is adopted to extract image features. ResNet adds the input and output of the unit together and then activates, on one hand, the speed of gradient disappearance is reduced, and on the other hand, small target defects can survive in a deep network for a longer time.
3.1 data set annotation
The prefabricated convex window mainly has 7 defect types of broken corners, pitted surfaces, honeycombs, cracks, air holes, slurry leakage and poor rough surfaces. Marking the collected sample data set, marking the defect type and the corresponding boundary box coordinate in the image data set by using a marking tool LabelImg to generate an xml file, compiling a Python script batch processing xml file, extracting the marking box coordinate and the target defect type to generate a txt file and a training set path file.
3.2 generating data sets
To generate the test set, 20% of the images from the annotated image dataset were selected as the test set, with approximately the same number of samples for each defect type, and the remainder were used to create the training set and validation set.
3.3 setting different size anchor points
In order to find out proper hyper-parameters by the model and improve the detection accuracy, anchor points with different scales and aspect ratios are set.
3.4 construction of fast R-CNN network model
A Petrel R-CNN convolutional neural network is built by utilizing a Pythrch frame to provide related convolution, pooling, an up-sampling function, a down-sampling function and the like, and the model is initialized.
3.5 Faster R-CNN model training
Training a fast RCNN model, training a marked prefabricated convex window sample and a corresponding label input model, iteratively updating by adopting a random gradient descent method, and comparing and selecting hyper-parameters until a loss function tends to be stable to finish training and storing model parameters.
3.6 fast R-CNN model test
And taking the prefabricated convex window test set as input to be loaded into the trained Faster R-CNN model, recording the test result and analyzing the optimization model.
And 4, step 4: defect determination 205 for prefabricated bay window
The optimized Faster R-CNN model is applied to detection and judgment of the viewing and sensing defects of the prefabricated convex window.
And 5: processing of detection results
And (4) the prefabricated convex window with the defect in the detection result enters a defect repairing process, and the prefabricated convex window without the defect enters the next production process.
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.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. A prefabricated convex window defect detection system based on improved Faster R-CNN comprises an image collection device (1) and a background PC (2), and is characterized in that: image collection device (1) includes lead screw (101), the equal threaded connection in both ends has slip nut (102) about lead screw (101) surface, the equal fixedly connected with extension rod (103) in both sides about slip nut (102), the surface of extension rod (103) is equipped with first fixed cover (104), the fixed cover of second (105) and the fixed cover of third (106) from being close to lead screw (101) to the direction of keeping away from lead screw (101) respectively, between first fixed cover (104) and the fixed cover of second (105) and between the fixed cover of second (105) and the fixed cover of third (106) and be located the equal fixedly connected with spring (107) in surface of extension rod (103), the equal fixedly connected with industry camera (108) in surface of first fixed cover (104), the fixed cover of second (105) and the fixed of third.
2. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 1, wherein: both ends are provided with the screw thread of equidirectional not about lead screw (101), the middle fixedly connected with gag lever post (109) on lead screw (101) surface, the left and right sides of gag lever post (109) is rotated and is connected with connecting rod (110), and one of them the bottom of connecting rod (110) is rotated with the top of the fixed cover of third (106) and is connected, another the bottom of connecting rod (110) is rotated with the top of the fixed cover of first (104) and is connected.
3. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 2, wherein: one fixed cover (105), a third fixed cover (106), two springs (107), three industry camera (108), a gag lever post (109) and two connecting rods (110) form a set of collection group extension rod (103), a first fixed cover (104), a second, image collection device (1) is provided with four sets of collection groups, and mutually perpendicular between two sets of adjacent collection groups.
4. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 1, wherein: the front side of lead screw (101) is provided with base plate (111), the rear side fixedly connected with servo motor (112) of base plate (111), the output and lead screw (101) fixed connection of servo motor (112).
5. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 4, wherein: the front side fixedly connected with support arm (113) of base plate (111), the bottom fixedly connected with base (114) of support arm (113), the bottom fixedly connected with pulley (115) of base (114).
6. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 1, wherein: the background PC (2) is fixed on the top of the base (114), and the background PC (2) comprises an analysis system (201).
7. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 6, wherein: the analysis system (201) comprises image preprocessing (202), improved Faster R-CNN (203), prefabricated convex window defect detection (204) and prefabricated convex window defect judgment (205).
8. The prefabricated bay window defect detection system based on improved Faster R-CNN as claimed in claim 7, wherein: the output end of the image preprocessing (202) is coupled with the input end of an improved Faster R-CNN (203), the output end of the improved Faster R-CNN (203) is coupled with the input end of a prefabricated convex window defect detection (204), and the output end of the prefabricated convex window defect detection (204) is coupled with the input end of a prefabricated convex window defect judgment (205).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011244564.1A CN112179846A (en) | 2020-11-10 | 2020-11-10 | Prefabricated convex window defect detection system based on improved Faster R-CNN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011244564.1A CN112179846A (en) | 2020-11-10 | 2020-11-10 | Prefabricated convex window defect detection system based on improved Faster R-CNN |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112179846A true CN112179846A (en) | 2021-01-05 |
Family
ID=73916584
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011244564.1A Pending CN112179846A (en) | 2020-11-10 | 2020-11-10 | Prefabricated convex window defect detection system based on improved Faster R-CNN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112179846A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113390875A (en) * | 2021-07-21 | 2021-09-14 | 华东交通大学 | Method for improving crack detection precision of steel fiber concrete |
-
2020
- 2020-11-10 CN CN202011244564.1A patent/CN112179846A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113390875A (en) * | 2021-07-21 | 2021-09-14 | 华东交通大学 | Method for improving crack detection precision of steel fiber concrete |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108562589B (en) | Method for detecting surface defects of magnetic circuit material | |
| Ali et al. | Structural crack detection using deep convolutional neural networks | |
| CN110314854B (en) | Workpiece detecting and sorting device and method based on visual robot | |
| CN108074231B (en) | Magnetic sheet surface defect detection method based on convolutional neural network | |
| Xue-Wu et al. | A vision inspection system for the surface defects of strongly reflected metal based on multi-class SVM | |
| CN110610475B (en) | Visual defect detection method of deep convolutional neural network | |
| CN107966454A (en) | A kind of end plug defect detecting device and detection method based on FPGA | |
| CN107486415A (en) | Thin bamboo strip defect on-line detecting system and detection method based on machine vision | |
| CN105891233A (en) | Intelligent detection system for surface defects of lens based on machine vision and implementation method of intelligent detection system | |
| CN111667455A (en) | AI detection method for various defects of brush | |
| CN113643268B (en) | Industrial product defect quality inspection method and device based on deep learning and storage medium | |
| CN111582294A (en) | Method for constructing convolutional neural network model for surface defect detection and application thereof | |
| CN107230203A (en) | Casting defect recognition methods based on human eye vision attention mechanism | |
| CN110473201A (en) | A kind of automatic testing method and device of disc surface defect | |
| CN106290392A (en) | A kind of little micro-bearing surface pitting defects online test method and system thereof | |
| CN109886932A (en) | Gear ring of wheel speed sensor detection method of surface flaw based on SVM | |
| Ji et al. | Apple grading method based on features of color and defect | |
| Zhao et al. | Defect detection method for electric multiple units key components based on deep learning | |
| Yao et al. | A feature memory rearrangement network for visual inspection of textured surface defects toward edge intelligent manufacturing | |
| Lin et al. | Development of a CNN-based hierarchical inspection system for detecting defects on electroluminescence images of single-crystal silicon photovoltaic modules | |
| Kähler et al. | Anomaly detection for industrial surface inspection: Application in maintenance of aircraft components | |
| Jin et al. | End Image Defect Detection of Float Glass Based on Faster Region-Based Convolutional Neural Network. | |
| CN112179846A (en) | Prefabricated convex window defect detection system based on improved Faster R-CNN | |
| Zhang et al. | Fabric defect detection based on visual saliency map and SVM | |
| Yang et al. | Crack identification of automobile steering knuckle fluorescent penetrant inspection based on deep convolutional generative adversarial networks data enhancement |
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 |