CN114723732A - Pleurotus eryngii surface flaw detection method and device and storage medium - Google Patents
Pleurotus eryngii surface flaw detection method and device and storage medium Download PDFInfo
- Publication number
- CN114723732A CN114723732A CN202210439052.3A CN202210439052A CN114723732A CN 114723732 A CN114723732 A CN 114723732A CN 202210439052 A CN202210439052 A CN 202210439052A CN 114723732 A CN114723732 A CN 114723732A
- Authority
- CN
- China
- Prior art keywords
- pleurotus eryngii
- image
- feature
- module
- network
- 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
- 244000252132 Pleurotus eryngii Species 0.000 title claims abstract description 153
- 235000001681 Pleurotus eryngii Nutrition 0.000 title claims abstract description 153
- 238000001514 detection method Methods 0.000 title claims abstract description 39
- 238000003860 storage Methods 0.000 title claims abstract description 8
- 230000007547 defect Effects 0.000 claims abstract description 22
- 238000000034 method Methods 0.000 claims abstract description 12
- 230000004927 fusion Effects 0.000 claims description 27
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000007246 mechanism Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 6
- 230000002950 deficient Effects 0.000 description 4
- 235000001674 Agaricus brunnescens Nutrition 0.000 description 3
- 230000002457 bidirectional effect Effects 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 241000222519 Agaricus bisporus Species 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000006116 polymerization reaction Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 241000222518 Agaricus Species 0.000 description 1
- 241000222350 Pleurotus Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000003786 synthesis 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
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a device for detecting surface flaws of pleurotus eryngii and a storage medium. The method comprises the following steps: s1: surrounding the pleurotus eryngii for at least one week, carrying out video acquisition, and analyzing the video into a pleurotus eryngii image frame sequence; s2: extracting a foreground target in the obtained pleurotus eryngii image frame; s3: constructing a condition generation countermeasure network comprising a generator and a discriminator, and respectively inputting the extracted foreground target into the generator and the discriminator to generate a standard pleurotus eryngii outline image; s4: extracting an original pleurotus eryngii outline image from a foreground target; s5: and performing sum and difference operation on the standard pleurotus eryngii outline image and the original pleurotus eryngii outline image to obtain the defect position of the pleurotus eryngii. According to the method, machine vision replaces manual detection, the detection efficiency of the pleurotus eryngii is improved, and flaw positions of pleurotus eryngii such as pleurotus eryngii with different shapes and changeable flaw shapes are detected by generating the standard pleurotus eryngii outline through the confrontation network.
Description
Technical Field
The invention belongs to the field of machine vision, and particularly relates to a pleurotus eryngii surface flaw detection method, a device and a storage medium.
Background
The intelligent detection of the surface flaws of the agricultural products gradually replaces the traditional manual detection and becomes a trend, which is a technology needing to be broken through in the traditional agricultural production enterprises in the advancing process of intelligent production.
At present, the method for detecting the flaws on the surfaces of edible mushrooms such as pleurotus eryngii and the like adopts a machine vision algorithm for detection besides manual detection. However, the pleurotus eryngii has different shapes and changeable flaw shapes, the algorithm designed manually is difficult to adapt to the changeable characteristics of the pleurotus eryngii, and meanwhile, the traditional machine vision detection algorithm has no generalization and is low in recognition accuracy. Therefore, an intelligent pleurotus eryngii surface flaw detection method based on deep learning is needed to be found, and flaw positions on the pleurotus eryngii surface can be identified quickly, flexibly and accurately.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a pleurotus eryngii surface flaw detection method which can quickly, flexibly and accurately identify the surface flaw positions of edible agaricus such as pleurotus eryngii.
Another objective of the present invention is to provide an apparatus capable of implementing the aforementioned pleurotus eryngii surface defect detection method, and a storage medium storing a computer program instantiated by the aforementioned pleurotus eryngii surface defect detection method.
The technical scheme is as follows: the pleurotus eryngii surface flaw detection method comprises the following steps:
s1: surrounding the pleurotus eryngii for at least one week, carrying out video acquisition, and analyzing the video into a pleurotus eryngii image frame sequence;
s2: extracting a foreground target in the obtained pleurotus eryngii image frame;
s3: constructing a condition generation countermeasure network comprising a generator and a discriminator, and respectively inputting the extracted foreground target into the generator and the discriminator to generate a standard pleurotus eryngii outline image;
s4: extracting an original pleurotus eryngii outline image from a foreground target;
s5: and performing sum and difference operation on the standard pleurotus eryngii outline image and the original pleurotus eryngii outline image to obtain the defect position of the pleurotus eryngii.
Further, the step S2 includes the steps of:
s2.1: constructing a target detection main feature network to extract a feature map in an image frame of pleurotus eryngii;
s2.2: acquiring an attention feature map of the feature map by adopting an attention algorithm;
s2.3: taking the obtained attention feature graph as node input, and obtaining fusion features by adopting a multi-scale feature fusion algorithm;
s2.4: inputting the fusion characteristics into a global pooling layer, and outputting the area and the category of the pleurotus eryngii in the image;
s2.5: and cutting out the pleurotus eryngii in the pleurotus eryngii image according to the area of the pleurotus eryngii in the image by adopting an image segmentation algorithm, and outputting a foreground target.
Further, in the step S2.3, the multi-scale feature fusion algorithm adopts a bidirectional fusion mode from top to bottom and from bottom to top.
Further, the target detection backbone feature network in step S2.1 is a MobileNet-SSD network.
The invention relates to a pleurotus eryngii surface flaw detection device which comprises a clamp, a driving mechanism, an image acquisition device and a processor, the clamp is used for clamping pleurotus eryngii, the driving mechanism is used for driving the clamp to rotate, the image acquisition device is used for carrying out video acquisition on the rotating pleurotus eryngii, the processor is used for calculating the flaw position and comprises a preprocessing module, a foreground object extraction module, a condition generation countermeasure module and a flaw position calculation module, the preprocessing module is used for analyzing the video into a pleurotus eryngii image frame sequence, the foreground object extracting module is used for extracting a pleurotus eryngii foreground object without a background from a pleurotus eryngii image frame, the condition generation countermeasure module is used for generating a standard pleurotus eryngii outline image and an original pleurotus eryngii outline image, and the defect position calculation module is used for performing sum and difference calculation on the pleurotus eryngii outline image and the original pleurotus eryngii outline image to obtain the defective position of the pleurotus eryngii.
Further, the device also comprises a background plate, and the clamp and the driving mechanism are arranged between the background plate and the image acquisition device.
Further, the foreground object extraction module comprises a main feature extraction network, a multi-channel attention submodule and a multi-scale feature fusion module, wherein the main feature extraction network is used for extracting a feature map in an pleurotus eryngii image frame, the multi-channel attention submodule is embedded in layers 6, 8, 10, 12 and 14 of the main feature extraction network and is used for acquiring an attention feature map from the feature map, and the multi-scale feature fusion module is used for acquiring fusion features from the attention feature map.
Further, the target detection backbone feature network is a MobileNet-SSD network.
Furthermore, the multi-scale feature fusion module adopts a bidirectional fusion mode from top to bottom and from bottom to top.
The storage medium of the present invention stores a computer program configured to implement the aforementioned pleurotus eryngii surface defect detection method when the computer program is run.
Has the advantages that: compared with the prior art, the invention has the following advantages: 1. the machine vision replaces manual detection, so that the detection efficiency of edible agaricus bisporus such as pleurotus eryngii is improved; 2. the generation countermeasure network is used for directly generating a standard pleurotus eryngii image from the original pleurotus eryngii image, so that defect positions of pleurotus eryngii with different forms and changeable defect shapes can be monitored; 3. the light-weight main network structure is used, the light-weight multi-channel attention module and the two-way fused multi-scale feature fusion module are added into the network, the detection precision and the monitoring speed are improved, meanwhile, the calculated amount is reduced, the network construction is convenient, the deployment difficulty is low, and the method is suitable for an actual processing site.
Drawings
FIG. 1 is a flow chart of a method for detecting surface defects of Pleurotus eryngii according to an embodiment of the present invention;
FIG. 2 is a mechanical structure diagram of the apparatus for detecting surface defects of Pleurotus eryngii according to the embodiment of the present invention;
FIG. 3 is a block diagram of a multi-channel attention module according to an embodiment of the present invention;
FIG. 4 is a block diagram of a multi-scale feature fusion module according to an embodiment of the present invention;
fig. 5 is a structural diagram of a target detection backbone feature network according to an embodiment of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings.
Referring to fig. 1, the pleurotus eryngii surface flaw detection method according to the embodiment of the invention comprises the following steps:
s1: surrounding the pleurotus eryngii for at least one week to collect videos, and analyzing the videos into a pleurotus eryngii image frame sequence;
s2: extracting a foreground target in the obtained pleurotus eryngii image frame;
s3: constructing a condition generation countermeasure network comprising a generator and a discriminator, and respectively inputting the extracted foreground target into the generator and the discriminator to generate a standard pleurotus eryngii outline image;
s4: extracting an original pleurotus eryngii outline image from a foreground target;
s5: and performing sum and difference operation on the standard pleurotus eryngii outline image and the original pleurotus eryngii outline image to obtain the defect position of the pleurotus eryngii.
According to the method for detecting the surface flaws of the pleurotus eryngii, machine vision replaces manual detection, and therefore detection efficiency of edible pleurotus eryngii such as pleurotus eryngii is improved. And the generation countermeasure network is adopted, the generator in the generation countermeasure network generates standard pleurotus eryngii outline images which are generated from the collected pleurotus eryngii image frames continuously and are identified by the identifier, and the standard pleurotus eryngii outline images and the identifier are optimized alternately, so that the generator can generate the standard pleurotus eryngii outline images, a large amount of time, manpower and material resources consumed for artificially manufacturing the standard pleurotus eryngii outline images are avoided, and meanwhile, the accuracy of the original pleurotus eryngii outline images extracted by the identifier is improved. And finally, performing sum and difference operation on the obtained standard pleurotus eryngii outline and the original pleurotus eryngii outline to obtain the surface defect positions of the pleurotus eryngii, and realizing the defect position detection of pleurotus eryngii such as pleurotus eryngii with different forms and changeable defect shapes.
In step S1, the pleurotus eryngii is clamped on the fixture 100, the fixture 100 is driven by the driving mechanism 200 to rotate, the image acquisition device 300 performs video shooting on the rotated pleurotus eryngii, images are captured at intervals of 15 frames to obtain a group of image frame sequences of the pleurotus eryngii, the outline of each pleurotus eryngii in each image frame sequence is calculated, and 12 image frames with the largest outline diameter are taken to build the target detection main feature network for extracting the foreground target.
Referring to fig. 4, in this embodiment, the sizes of convolution kernels of the network convolution layers of the trunk feature extraction network are 3 × 3 and 5 × 5, respectively, and a lightweight multi-channel attention module is embedded after the 6 th, 8 th, 10 th, 12 th and 14 th layers of the trunk network, and a multi-channel feature fusion module is disposed between the multi-channel attention module and the global average pooling layer of the trunk feature extraction network, and a specific foreground object extraction process is as shown in fig. 3:
step 2.1: inputting 12 image frames of the pleurotus eryngii selected in the step 1 into a main feature extraction network, and outputting feature maps X of layers 6, 8, 10, 12 and 14 of the main feature extraction networkiInputting the data into a global average pooling layer of a backbone network for feature aggregation to obtain an aggregated feature yiWherein the characteristic diagram XiThe sizes are respectively C × Hi×WiPolymerization characteristics yiThe size is C × 1 × 1, and the formula is:
wherein ε is the weight coefficient, which in this example is 0.4.
Step 2.2: the polymerization characteristics y obtained in step 2.1iInputting the local cross-channel information into a 1 multiplied by 1 convolution kernel of the multi-channel attention module to carry out local cross-channel information interaction to obtain an interaction feature map omegaiAnd the output is output after the action of the activation function sigma, and the formula is as follows:
ωi＝σ(C1Dk(yi))(i＝6、8、10、12、14) (2)
step 2.3: will feature diagram XiInteraction feature map omegaiPerforming element product to obtain cross-channel attention feature map
Trans-channel attentiveness featureSign graph
Has a size of CxHi×WiThe formula is as follows:
step 2.4: inputting the cross-channel attention feature map output in the step 2.3 into a multi-scale feature fusion algorithm module as 5 feature nodes, inputting the feature nodes into a global average pooling layer after feature fusion, and outputting the region and category of the pleurotus eryngii in the image frame;
step 2.5: and (3) cutting out an pleurotus eryngii foreground target according to the area of the pleurotus eryngii in the image by using an image segmentation algorithm, and outputting the pleurotus eryngii image without the background.
In order to reduce the calculation amount, lighten the backbone network structure, reduce the deployment difficulty, and be more suitable for actual processing field operation, in this embodiment, the target detection backbone network adopts a MobileNet-SSD network, and adopts a deep separable convolution to perform decomposition calculation on a standard convolution kernel, thereby reducing the calculation amount. Meanwhile, the multi-channel feature fusion module adopts a bidirectional fusion mode from top to bottom and from bottom to top, a shortcut is added between the bottommost layer and the highest layer, the path between the layers is shortened, and a specific synthesis mode can be described as follows:
step 2.4.1: fusing the characteristic nodes 6 and 8 to obtain a node A, wherein A is 6+ 8;
step 2.4.2: fusing the characteristic nodes A and 10 to obtain a node B, wherein the node B is A + 10;
step 2.4.3: fusing the characteristic nodes B and 12 to obtain a node C, wherein the C is B + 12;
step 2.4.4: fusing the characteristic nodes C and 14 to obtain a node D, wherein D is C + 14;
step 2.4.5: fusing the characteristic point D with C and 12 to obtain a node E, wherein E is D + C + 12;
step 2.4.6: fusing the characteristic point E with the characteristic points B and 10 to obtain a node F, wherein F is E + B + 10;
step 2.4.7: fusing the characteristic point F with the A and the 8 to obtain a node G, wherein G is F + A + 8;
step 2.4.8: fusing the characteristic points G and 6 to obtain a node H, wherein G is H + 6;
step 2.4.9: the feature node D, E, F, G, H is input into the global pooling layer and output as the area and category of Pleurotus eryngii in the image.
In the embodiment, a GrabCut algorithm is adopted, a foreground and a background are modeled according to the position of the pleurotus eryngii in an image frame predicted by an input target detection backbone network, segmentation is realized by adopting a minCut algorithm, and a foreground target image which is free of the background and only keeps the foreground of the pleurotus eryngii is output.
In the present embodiment, the generator network employs a UNet network, and the discriminator network employs a patchGAN network. The generator generates a standard pleurotus eryngii image according to the pleurotus eryngii foreground target image, inputs the standard pleurotus eryngii image into the discriminator, extracts the standard pleurotus eryngii image characteristics, and calculates the probability distribution P of the image characteristicsnormPerforming identification when probability distribution PnormProbability distribution P of pleurotus eryngii approaching to real standardTureAnd then, the generator outputs the outline image of the standard pleurotus eryngii image, and performs sum and difference operation with the outline image of the pleurotus eryngii foreground target image, namely the original pleurotus eryngii outline image, so as to obtain the defective position of the pleurotus eryngii.
In the process of extracting the profile of the pleurotus eryngii, firstly converting a pleurotus eryngii foreground target image into a gray-scale image, denoising the image through a Gaussian filtering algorithm, finally extracting the pleurotus eryngii profile in the image by adopting a Canny algorithm, and outputting an original pleurotus eryngii profile image.
In this embodiment, defective Pixel point Pixel of Pleurotus eryngiiiThe calculation method of (2) is as follows:
ISUMi＝Ii+Pi (4)
Pixeli＝ISUMi-Pi (5)
wherein ISUMiIs an original pleurotus eryngii outline image Ii(i-1 … 12) and a standard Pleurotus eryngii silhouette image Pi(i-1 … 12) is added according to an image algorithm. Extracting defective Pixel points PixeliCoordinate of (c) [ (u)1,v1),…,(un,vn)]And obtaining the position coordinates of the defect of the pleurotus eryngii.
It is understood that the above method can also be applied to edible agaricus bisporus, bolete and the like with similar structures.
Referring to fig. 2, the apparatus for detecting surface defects of pleurotus eryngii according to the embodiment of the present invention includes a fixture 100, a driving mechanism 200, an image capturing device 300, and a processor 400. The fixture 100 is used for clamping pleurotus eryngii, the driving mechanism 200 is used for driving the fixture 100 to rotate, the image acquisition device 300 is used for carrying out video acquisition on the rotated pleurotus eryngii, and the processor 400 is used for calculating the defect position. The processor 400 comprises a preprocessing module, a foreground object extraction module, a condition generation countermeasure module and a flaw position calculation module, wherein the preprocessing module is used for analyzing a video into a pleurotus eryngii image frame sequence, the foreground object extraction module is used for extracting a pleurotus eryngii foreground object without a background from a pleurotus eryngii image frame, the condition generation countermeasure module is used for generating a standard pleurotus eryngii contour image and an original pleurotus eryngii contour image, and the flaw position calculation module is used for performing sum and difference calculation on the pleurotus eryngii contour image and the original pleurotus eryngii contour image to obtain a pleurotus eryngii flaw position.
In the present embodiment, in order to further reduce the amount of calculation, a monochrome background plate 500 is further provided, and the jig 100 and the driving mechanism 200 are provided between the background plate 500 and the image pickup device 300. Meanwhile, in order to ensure that the pleurotus eryngii clamped by the clamp 100 rotates at a constant speed, the driving mechanism 200 comprises a servo motor and a gear set, and the servo motor drives the clamp 100 to rotate through the gear set formed by the large pinion. The storage medium of the embodiment of the invention stores the computer program instantiated by the pleurotus eryngii surface defect detection method.
Claims (10)
1. A pleurotus eryngii surface flaw detection method is characterized by comprising the following steps:
s1: surrounding the pleurotus eryngii for at least one week to collect videos, and analyzing the videos into a pleurotus eryngii image frame sequence;
s2: extracting a foreground target in the obtained pleurotus eryngii image frame;
s3: constructing a condition generation countermeasure network comprising a generator and a discriminator, and respectively inputting the extracted foreground target into the generator and the discriminator to generate a standard pleurotus eryngii outline image;
s4: extracting an original pleurotus eryngii outline image from a foreground target;
s5: and performing sum and difference operation on the standard pleurotus eryngii outline image and the original pleurotus eryngii outline image to obtain the defect position of the pleurotus eryngii.
2. The pleurotus eryngii surface defect detecting method according to claim 1, wherein the step S2 includes the steps of:
s2.1: constructing a target detection main feature network to extract a feature map in an image frame of pleurotus eryngii;
s2.2: acquiring an attention feature map of the feature map by adopting an attention algorithm;
s2.3: taking the obtained attention feature graph as a node input, and obtaining a fusion feature by adopting a multi-scale feature fusion algorithm;
s2.4: inputting the fusion characteristics into a global pooling layer, and outputting the area and the category of the pleurotus eryngii in the image;
s2.5: and cutting out the pleurotus eryngii in the pleurotus eryngii image according to the area of the pleurotus eryngii in the image by adopting an image segmentation algorithm, and outputting a foreground object.
3. The pleurotus eryngii surface blemish detection method according to claim 2, wherein in step S2.3, the multi-scale feature fusion algorithm employs a top-down and bottom-up bi-directional fusion mode.
4. The method for detecting the surface flaws of Pleurotus eryngii according to claim 2, wherein the target detection backbone feature network in step S2.1 is a MobileNet-SSD network.
5. The pleurotus eryngii surface flaw detection device is characterized by comprising a clamp, a driving mechanism, an image acquisition device and a processor, wherein the clamp is used for clamping pleurotus eryngii, the driving mechanism is used for driving the clamp to rotate, the image acquisition device is used for carrying out video acquisition on the rotating pleurotus eryngii, the processor is used for calculating flaw positions, the processor comprises a preprocessing module, a foreground object extraction module, a condition generation countermeasure module and a flaw position calculation module, the preprocessing module is used for analyzing a video into a pleurotus eryngii image frame sequence, the foreground object extraction module is used for extracting a background-free pleurotus eryngii foreground object from a pleurotus eryngii image frame, the condition generation countermeasure module is used for generating a standard pleurotus eryngii contour image and an original pleurotus eryngii contour image, and the flaw position calculation module is used for carrying out sum-difference calculation on the pleurotus eryngii contour image and the original pleurotus eryngii contour image to obtain the pleurotus eryngii flaw positions.
6. The apparatus of claim 5, further comprising a background plate, wherein the clamp and the driving mechanism are disposed between the background plate and the image capturing device.
7. The apparatus according to claim 5, wherein the foreground object extraction module comprises a main feature extraction network, a multi-channel attention submodule and a multi-scale feature fusion module, the main feature extraction network is used for extracting a feature map in an image frame of Pleurotus eryngii, the multi-channel attention submodule is embedded behind layers 6, 8, 10, 12 and 14 of the main feature extraction network and is used for acquiring an attention feature map from the feature map, and the multi-scale feature fusion module is used for acquiring a fusion feature from the attention feature map.
8. The apparatus of claim 7, wherein the target detection backbone feature network is a MobileNet-SSD network.
9. The pleurotus eryngii surface defect detecting device according to claim 7, wherein the multi-scale feature fusion module adopts a top-down and bottom-up two-way fusion mode.
10. A storage medium storing a computer program, wherein the computer program is configured to implement the Pleurotus eryngii defect surface detection method according to any one of claims 1-4 when the computer program is run.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210439052.3A CN114723732A (en) | 2022-04-25 | 2022-04-25 | Pleurotus eryngii surface flaw detection method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210439052.3A CN114723732A (en) | 2022-04-25 | 2022-04-25 | Pleurotus eryngii surface flaw detection method and device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114723732A true CN114723732A (en) | 2022-07-08 |
Family
ID=82244923
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210439052.3A Pending CN114723732A (en) | 2022-04-25 | 2022-04-25 | Pleurotus eryngii surface flaw detection method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114723732A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110434090A (en) * | 2019-08-23 | 2019-11-12 | 中山德著智能科技有限公司 | A kind of carrot surface flaw detecting method |
| CN111445484A (en) * | 2020-04-01 | 2020-07-24 | 华中科技大学 | Image-level labeling-based industrial image abnormal area pixel level segmentation method |
| CN113610848A (en) * | 2021-10-09 | 2021-11-05 | 阿里巴巴（中国）有限公司 | Digital cloth processing system, cloth flaw detection method, device and medium |
| CN113902901A (en) * | 2021-09-30 | 2022-01-07 | 北京工业大学 | Object separation method and system based on lightweight detection |
| CN113989662A (en) * | 2021-10-18 | 2022-01-28 | 中国电子科技集团公司第五十二研究所 | Remote sensing image fine-grained target identification method based on self-supervision mechanism |
-
2022
- 2022-04-25 CN CN202210439052.3A patent/CN114723732A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110434090A (en) * | 2019-08-23 | 2019-11-12 | 中山德著智能科技有限公司 | A kind of carrot surface flaw detecting method |
| CN111445484A (en) * | 2020-04-01 | 2020-07-24 | 华中科技大学 | Image-level labeling-based industrial image abnormal area pixel level segmentation method |
| CN113902901A (en) * | 2021-09-30 | 2022-01-07 | 北京工业大学 | Object separation method and system based on lightweight detection |
| CN113610848A (en) * | 2021-10-09 | 2021-11-05 | 阿里巴巴（中国）有限公司 | Digital cloth processing system, cloth flaw detection method, device and medium |
| CN113989662A (en) * | 2021-10-18 | 2022-01-28 | 中国电子科技集团公司第五十二研究所 | Remote sensing image fine-grained target identification method based on self-supervision mechanism |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113674247B (en) | X-ray weld defect detection method based on convolutional neural network | |
| CN111007073B (en) | Method and system for online detection of part defects in additive manufacturing process | |
| CN111080693A (en) | Robot autonomous classification grabbing method based on YOLOv3 | |
| CN108460764A (en) | The ultrasonoscopy intelligent scissor method enhanced based on automatic context and data | |
| CN111401293B (en) | Gesture recognition method based on Head lightweight Mask scanning R-CNN | |
| US20220366682A1 (en) | Computer-implemented arrangements for processing image having article of interest | |
| CN115439458A (en) | Industrial image defect target detection algorithm based on depth map attention | |
| CN113393426A (en) | Method for detecting surface defects of rolled steel plate | |
| CN111160356A (en) | Image segmentation and classification method and device | |
| CN114155474A (en) | Damage identification technology based on video semantic segmentation algorithm | |
| CN113111875A (en) | Seamless steel rail weld defect identification device and method based on deep learning | |
| CN115908354A (en) | Photovoltaic panel defect detection method based on double-scale strategy and improved YOLOV5 network | |
| CN114037684B (en) | Defect detection method based on yolov and attention mechanism model | |
| CN116703919A (en) | Surface impurity detection method based on optimal transmission distance loss model | |
| CN115049945A (en) | Method and device for extracting lodging area of wheat based on unmanned aerial vehicle image | |
| CN115115917A (en) | 3D point cloud target detection method based on attention mechanism and image feature fusion | |
| CN114332083A (en) | PFNet-based industrial product camouflage flaw identification method | |
| Wang et al. | MeDERT: A metal surface defect detection model | |
| CN114723732A (en) | Pleurotus eryngii surface flaw detection method and device and storage medium | |
| Kumar et al. | Self supervised learning for multiple object tracking in 3d point clouds | |
| CN115082787A (en) | Cascade R-CNN-based target detection method for fruits to be picked | |
| CN115019202A (en) | Step-by-step grabbing detection method applied to service type mobile mechanical arm | |
| Ullah et al. | Real-time video dehazing for industrial image processing | |
| Kent et al. | Deep Learning Approach for Detection of Oil Palm Tree on UAV Images | |
| CN115578295B (en) | Video rain removing method, system, equipment and storage medium |
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 |