CN111136027A - Salted duck egg quality sorting device and method based on convolutional neural network - Google Patents
Salted duck egg quality sorting device and method based on convolutional neural network Download PDFInfo
- Publication number
- CN111136027A CN111136027A CN202010037406.2A CN202010037406A CN111136027A CN 111136027 A CN111136027 A CN 111136027A CN 202010037406 A CN202010037406 A CN 202010037406A CN 111136027 A CN111136027 A CN 111136027A
- Authority
- CN
- China
- Prior art keywords
- salted duck
- conveying mechanism
- duck eggs
- neural network
- convolutional neural
- 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.)
- Granted
Links
- 241000272525 Anas platyrhynchos Species 0.000 title claims abstract description 237
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 23
- 235000013601 eggs Nutrition 0.000 claims abstract description 248
- 238000001514 detection method Methods 0.000 claims abstract description 65
- 230000007246 mechanism Effects 0.000 claims description 171
- 239000007789 gas Substances 0.000 claims description 45
- QGZKDVFQNNGYKY-UHFFFAOYSA-N Ammonia Chemical compound N QGZKDVFQNNGYKY-UHFFFAOYSA-N 0.000 claims description 35
- RWSOTUBLDIXVET-UHFFFAOYSA-N Dihydrogen sulfide Chemical compound S RWSOTUBLDIXVET-UHFFFAOYSA-N 0.000 claims description 35
- 229910000037 hydrogen sulfide Inorganic materials 0.000 claims description 35
- 238000012549 training Methods 0.000 claims description 14
- 230000001360 synchronised effect Effects 0.000 claims description 13
- 238000011176 pooling Methods 0.000 claims description 12
- 210000003278 egg shell Anatomy 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 6
- 238000007477 logistic regression Methods 0.000 claims description 6
- 101150021123 msrA gene Proteins 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 230000001502 supplementing effect Effects 0.000 claims description 5
- 230000008901 benefit Effects 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000007664 blowing Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 abstract description 3
- 244000144977 poultry Species 0.000 description 6
- 210000000078 claw Anatomy 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000013589 supplement Substances 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 235000016709 nutrition Nutrition 0.000 description 3
- 230000035764 nutrition Effects 0.000 description 3
- 239000000047 product Substances 0.000 description 3
- 102000002322 Egg Proteins Human genes 0.000 description 2
- 108010000912 Egg Proteins Proteins 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 210000001503 joint Anatomy 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- TVEXGJYMHHTVKP-UHFFFAOYSA-N 6-oxabicyclo[3.2.1]oct-3-en-7-one Chemical compound C1C2C(=O)OC1C=CC2 TVEXGJYMHHTVKP-UHFFFAOYSA-N 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005554 pickling Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 102000004169 proteins and genes Human genes 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000011573 trace mineral Substances 0.000 description 1
- 235000013619 trace mineral Nutrition 0.000 description 1
- 235000021404 traditional food Nutrition 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
- 239000011782 vitamin Substances 0.000 description 1
- 229940088594 vitamin Drugs 0.000 description 1
- 229930003231 vitamin Natural products 0.000 description 1
- 235000013343 vitamin Nutrition 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K43/00—Testing, sorting or cleaning eggs ; Conveying devices ; Pick-up devices
- A01K43/04—Grading eggs
- A01K43/10—Grading and stamping
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/02—Measures preceding sorting, e.g. arranging articles in a stream orientating
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
- B07C5/342—Sorting according to other particular properties according to optical properties, e.g. colour
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/361—Processing or control devices therefor, e.g. escort memory
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/361—Processing or control devices therefor, e.g. escort memory
- B07C5/362—Separating or distributor mechanisms
-
- 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
-
- 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
- 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
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/0004—Gaseous mixtures, e.g. polluted air
- G01N33/0009—General constructional details of gas analysers, e.g. portable test equipment
- G01N33/0027—General constructional details of gas analysers, e.g. portable test equipment concerning the detector
- G01N33/0036—General constructional details of gas analysers, e.g. portable test equipment concerning the detector specially adapted to detect a particular component
- G01N33/0044—Sulphides, e.g. H2S
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/02—Food
- G01N33/08—Eggs, e.g. by candling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
-
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C2501/00—Sorting according to a characteristic or feature of the articles or material to be sorted
- B07C2501/0081—Sorting of food items
-
- 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
- G01N2021/8883—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 involving the calculation of gauges, generating models
Landscapes
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Food Science & Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Environmental Sciences (AREA)
- Medicinal Chemistry (AREA)
- Combustion & Propulsion (AREA)
- Sorting Of Articles (AREA)
- Biodiversity & Conservation Biology (AREA)
- Animal Husbandry (AREA)
- Neurology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Signal Processing (AREA)
Abstract
The invention discloses a salted duck egg quality sorting device and method based on a convolutional neural network, wherein the salted duck egg quality sorting device comprises a rack, a black box arranged on the rack, a detection device arranged in the black box, a conveying device used for conveying salted duck eggs into the black box for detection and a control system; according to the salted duck egg quality sorting device, smelly salted duck eggs are preferentially removed by detecting gas components in gas emitted by the salted duck eggs, the remaining salted duck eggs are detected by using the convolutional neural network model which is trained in advance and used for detecting the salted duck egg quality, the inferior salted duck eggs and the high-quality salted duck eggs are classified, the traditional manual detection mode is replaced, the labor force is saved, and the production efficiency is improved.
Description
Technical Field
The invention relates to salted duck egg detection equipment, in particular to a salted duck egg quality sorting device and method based on a convolutional neural network.
Background
Salted duck eggs are also called salted duck eggs and are a traditional food in China. The salted duck egg is a reproduced egg prepared by pickling fresh duck eggs serving as main raw materials, is rich in nutrition, is rich in fat, protein and various trace elements, vitamins and the like required by a human body, and is easy to absorb by the human body. The high-quality salted duck eggs are mellow and smooth in appearance, the eggshells are complete and clean, no cracks exist, and the shells are cyan; the bad salted duck eggs have grey shells, more white or black spots on the surfaces, thin and fragile eggshells and easy breakage, even the deteriorated salted duck eggs can generate pungent odor which is mainly derived from chemical gases such as ammonia gas, hydrogen sulfide gas and the like, and because a plurality of tiny air holes are formed on the surfaces of the eggshells, even the salted duck eggs with complete shells can still emit odor.
Poultry eggs such as eggs and duck eggs and egg products are important ways for guaranteeing sufficient nutrition of people, and various grading and detection technologies for poultry eggs are developed, but most of the poultry eggs are researched for fresh eggs, most of the poultry eggs are used for detecting nutrition or harmful ingredients of the poultry eggs, and the poultry eggs and the egg products prepared by processing the salted duck eggs in a certain way are mainly used for detecting the quality and the quality of the salted duck eggs and still are mainly selected and classified manually, so that the method is time-consuming, labor-consuming, low in efficiency and easy to make mistakes. Therefore, devices and methods for detecting and analyzing the quality of salted duck eggs appear in the market, for example, the utility model patent with application publication number CN108051449A discloses an "online visual detection method for surface cracks of salted duck eggs based on morphological edge detection", which can realize online detection and classification of cracks of salted duck eggs by using machine vision for detection, and has low requirements on application environment, low equipment cost and better application and popularization prospects compared with acoustic detection; a single industrial camera is used for photographing salted duck eggs at multiple turnover angles, and the detection range is full; and the cracks are extracted by utilizing morphological edge detection, so that the method is more visual and accurate.
However, the online visual detection method for the surface cracks of the salted duck eggs has the following defects:
although the salted duck eggs which are likely to deteriorate are removed by detecting the surface cracks of the salted duck eggs, the inferior salted duck egg shells are still not cracked but have been deteriorated to different degrees in the market, so that the online visual detection method for the surface cracks of the salted duck eggs is difficult to detect.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides the salted duck egg quality sorting device based on the convolutional neural network, the salted duck egg quality sorting device can be used for more accurately detecting the quality of salted duck eggs, and the detection precision is higher.
The invention also aims to provide a salted duck egg quality sorting method based on the convolutional neural network.
The technical scheme for solving the technical problems is as follows:
a salted duck egg quality sorting device based on a convolutional neural network comprises a rack, a black box arranged on the rack, a detection device arranged in the black box, a conveying device used for conveying salted duck eggs into the black box for detection and a control system, wherein,
two sides of the black box are respectively provided with a salted duck egg inlet and a salted duck egg outlet, wherein the salted duck egg inlet is provided with an inlet baffle and a first driving motor for driving the inlet baffle to rotate so as to enable salted duck eggs to enter the black box one by one, the first driving motor is installed on the black box, and a main shaft of the first driving motor is connected with the inlet baffle;
the conveying device comprises a first conveying mechanism, a second conveying mechanism and a third conveying mechanism, wherein the second conveying mechanism penetrates through the black box, the head end of the second conveying mechanism is connected with the tail end of the first conveying mechanism, and the tail end of the second conveying mechanism is connected with the head end of the third conveying mechanism; the third conveying mechanism is used for conveying the detected salted duck eggs to different collecting mechanisms;
the detection device comprises a first detection module arranged in the black box and a second detection module used for detecting the surface of the salted duck eggs, wherein,
the first detection module comprises an ammonia gas sensor, a hydrogen sulfide gas sensor and a fan which are arranged in a black box, wherein the ammonia gas sensor and the hydrogen sulfide gas sensor are arranged on the inner wall of the black box and are positioned on one side of the second conveying mechanism and used for detecting whether ammonia gas and hydrogen sulfide gas are contained in the gas emitted by salted duck eggs; the fan is arranged on the inner wall of the black box, is arranged opposite to the ammonia gas sensor and the hydrogen sulfide gas sensor, and is used for blowing gas emitted by salted duck eggs into the detection range of the ammonia gas sensor and the hydrogen sulfide gas sensor;
the second detection module comprises an image collection device, and is used for collecting the images of the salted duck eggs detected in the black box, transmitting the images to the control system, and analyzing and processing the images by the control system.
Preferably, the image collecting device comprises an industrial camera, and the industrial camera is arranged on the top of the inner wall of the black box and is positioned above the second conveying mechanism.
Preferably, the image collecting device further comprises light supplement lamps, and the light supplement lamps are installed on two sides of the industrial camera and used for supplementing light.
Preferably, an infrared sensor is arranged between the ammonia gas sensor and the hydrogen sulfide gas sensor and used for detecting whether salted duck eggs are detected.
Preferably, an inner baffle and a second driving motor for driving the inner baffle to rotate are further arranged in the black box, wherein the inner baffle is located above the second conveying mechanism; the second driving motor is installed on the inner wall of the black box, and a main shaft of the second driving motor is connected with the inner baffle.
Preferably, the conveying speed of the first conveying mechanism is lower than that of the second conveying mechanism.
Preferably, the number of the third conveying mechanisms is multiple, the head end of each group of the third conveying mechanisms is communicated with the tail end of the second conveying mechanism, and the tail ends of the third conveying mechanisms are connected with the collecting mechanisms; the end of the second conveying mechanism is provided with a sorting baffle, and the sorting baffle is driven by a third driving motor and is used for conveying salted duck eggs into a corresponding third conveying mechanism.
Preferably, the first conveying mechanism, the second conveying mechanism and the third conveying mechanism are all synchronous belt transmission mechanisms.
A salted duck egg quality sorting method based on a convolutional neural network comprises the following steps:
(1) firstly, the first conveying mechanism drives salted duck eggs to move, the salted duck eggs are conveyed to the second conveying mechanism, and then the salted duck eggs are conveyed to the detection range of the detection device by the second conveying mechanism; then, the first driving motor drives the inlet baffle to move, and the salted duck egg inlet in the black box is blocked;
(2) the control system controls the fan to work, the odor of the salted duck eggs is blown to the ammonia gas sensor and the hydrogen sulfide gas sensor, and the detection information is transmitted to the control system by the ammonia gas sensor and the hydrogen sulfide gas sensor; if the detected contents of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs exceed the standard values, the salted duck eggs are unqualified; the second conveying mechanism conveys the salted duck eggs to a third conveying mechanism, and the salted duck eggs are conveyed to a collecting mechanism for collecting unqualified salted duck eggs through the third conveying mechanism;
(3) if the detected content of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs does not exceed the standard value, the control system controls the image collecting device to work, photographs of the salted duck eggs are shot, and the photograph information is sent to the control system; the control system constructs a convolutional neural network model by processing the picture, wherein the construction of the convolutional neural network model comprises the following steps:
(3-1) collecting sample images and establishing a sample library, wherein the collected sample images comprise color photographs of salted duck eggs with different advantages and disadvantages taken at different angles; carrying out category marking on the obtained photos of the salted duck eggs, and the method comprises the following steps: the salted duck eggs with round and smooth appearance, complete and clean eggshells, no cracks and cyan shells are marked as positive type; marking salted duck eggs with dark shells, more white or black spots on the surfaces or cracks on the surfaces as reverse categories; if the input pictures contain pictures which do not contain salted duck eggs, marking the pictures as irrelevant pictures; then saving 70% of the labeled images as a training sample set and 30% as a testing sample set;
(3-2) normalizing the input sample image, and normalizing the input sample image in the training sample set and the test sample set to 227 multiplied by 227 for subsequent input into the convolutional neural network;
(3-3) building a convolutional neural network model; the constructed convolutional neural network model is an AlexNet model, and the network layer of the convolutional neural network model comprises 5 convolutional layers, 3 maximum pooling layers, 3 full-connection layers and 1 output layer; wherein, the first convolution layer adopts convolution kernel with the size of 11 multiplied by 11 and the step length is 4, and then a maximum pooling layer with the size of 3 multiplied by 3 is followed and the step length is 2; the second convolution layer adopts convolution kernel with the size of 5 multiplied by 5, the image size is kept unchanged by adopting image filling operation, and then a maximum pooling layer with the size of 3 multiplied by 3 is followed, and the step length is 2; subsequently, the third, fourth and fifth convolutional layers are three tightly connected identical convolutional layers of size 3 × 3, and the image size is kept unchanged by image filling operation; next to a maximum pooling layer of size 3 × 3, the step size is 2; followed by a fully-connected layer with an output of 9216 dimensions, followed by two identical fully-connected layers with an output of 4096 dimensions, followed by the output layer; the output layer adopts a Softmax function as an excitation function, and the rest network layers adopt a Relu function as an excitation function;
(3-4) initializing a convolution neural network model, initializing by using an Msra algorithm, and when only the input number is considered, initializing the Msra to be Gaussian distribution with the mean value of 0 and the variance of 2/n;
(3-5) training a convolutional neural network model;
(3-6) classifying the salted duck eggs through a trained convolutional neural network model; classifying the normally shot pictures containing the salted duck eggs into a positive type or a negative type; classifying the pictures of the non-salted duck eggs shot by mistake into irrelevant classes;
(4) and the control system controls the third conveying mechanism to convey the salted duck eggs after detection to the corresponding collecting mechanism according to the classification result of the convolutional neural network model.
Preferably, in step (3-5), the step of training the convolutional neural network model is:
(3-51), inputting characteristic variables;
(3-52) carrying out logic calculation on each item of data in the characteristic variables, and adding a normalization item during calculation;
(3-53) calculating the output of each neuron of each layer, and randomly abandoning the output of a part of neurons in the first two fully-connected layers by adopting a Dropout method;
(3-54) calculating a cost function of the logistic regression;
(3-55) calculating a proper weight matrix and a proper bias vector value by using a gradient descent method, and minimizing a cost function of logistic regression;
(3-56) repeating the steps (3-53) - (3-55) until the prediction accuracy meets the requirement;
and (3-57) analyzing and outputting results.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the method, smelly salted duck eggs are preferentially removed by detecting gas components in gas emitted by the salted duck eggs, the remaining salted duck eggs are detected by utilizing the convolution neural network model which is trained in advance and used for detecting the quality of the salted duck eggs, the inferior salted duck eggs and the high-quality salted duck eggs are classified, the traditional manual detection mode is replaced, the labor force is saved, and the production efficiency is improved.
2. According to the invention, the salted duck eggs which are seriously deteriorated are preferentially removed by detecting the gas components emitted by the salted duck eggs, the remaining salted duck eggs are detected by using the convolutional neural network, the surface spots can still be used as the basis for classifying the salted duck eggs with complete shells, and finally the high-quality salted duck eggs and the low-quality salted duck eggs can be classified with higher accuracy, so that the detection precision is improved.
3. The method can be applied to the product quality detection link before salted duck eggs are produced and packaged, replaces the traditional manual detection classification, improves the production efficiency, and effectively controls the outgoing quality of the salted duck eggs.
Drawings
Fig. 1 is a schematic structural diagram of a salted duck egg quality sorting device based on a convolutional neural network according to a first embodiment of the invention.
FIG. 2 is a schematic view of the black box, the detecting device and the second conveying mechanism.
Fig. 3 is a schematic flow chart of the salted duck egg quality sorting method based on the convolutional neural network.
Fig. 4 is a schematic flow chart of the convolutional neural network model constructed in fig. 3.
Fig. 5 is a schematic flow chart of the process of training the convolutional neural network model in fig. 4.
Fig. 6 is a schematic diagram of a network structure of a convolutional neural network model.
Fig. 7 is a schematic configuration diagram of the first conveying mechanism and the second conveying mechanism in the salted duck egg quality sorting device based on the convolutional neural network according to the second embodiment of the present invention.
Fig. 8 is a schematic configuration diagram of a salted duck egg quality sorting device based on a convolutional neural network according to a third embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.
Example 1
Referring to fig. 1-2, the salted duck egg quality sorting device based on the convolutional neural network comprises a rack, a black box 7 arranged on the rack, a detection device arranged in the black box 7, a conveying device 11 for conveying salted duck eggs 8 into the black box 7 for detection, and a control system.
Referring to fig. 1-2, a salted duck egg inlet and a salted duck egg outlet are respectively arranged on two sides of the black box 7, wherein the salted duck egg inlet is provided with an inlet baffle 9 and a first driving motor for driving the inlet baffle 9 to rotate so as to promote salted duck eggs 8 to enter the black box 7 one by one, the first driving motor is mounted on the black box 7, and a main shaft is connected with the inlet baffle 9.
Referring to fig. 1-2, the conveying device 11 comprises a first conveying mechanism 11-1, a second conveying mechanism 11-2 and a third conveying mechanism 11-3, wherein the second conveying mechanism 11-2 penetrates through the black box 7, the head end of the second conveying mechanism 11-2 is connected with the tail end of the first conveying mechanism 11-1, and the tail end is connected with the head end of the third conveying mechanism 11-3; the third conveying mechanism 11-3 is used for conveying the salted duck eggs 8 after detection to different collecting mechanisms 11-4.
Referring to fig. 1-2, the detection device comprises a first detection module arranged in a black box 7 and a second detection module used for detecting the surface of the salted duck egg 8, wherein the first detection module comprises an ammonia gas sensor 3, a hydrogen sulfide gas sensor 5 and a fan 13 which are arranged in the black box 7, the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5 are mounted on the inner wall of the black box 7 and are positioned at the side edge of the second conveying mechanism 11-2 and used for detecting whether ammonia gas and hydrogen sulfide gas are contained in the gas emitted by the salted duck egg 8 or not; the fan is arranged on the inner wall of the black box 7, is arranged opposite to the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5, and is used for blowing gas emitted by the salted duck eggs 8 into the detection range of the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5; an infrared sensor 4 is arranged between the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5 and is used for detecting whether salted duck eggs 8 are detected;
the second detection module comprises an image collection device, the image collection device is used for collecting images of salted duck eggs 8 detected in a black box 7, transmitting the images to a control system, and analyzing and processing the images by the control system, wherein the image collection device comprises an industrial camera 1 and a light supplement lamp 2, the industrial camera 1 is mounted on the top of the inner wall of the black box 7 through a camera support 6 and is positioned above the second conveying mechanism 11-2; the light supplement lamps 2 are installed on two sides of the industrial camera 1 and used for supplementing light.
Referring to fig. 1-2, an inner baffle 10 and a second driving motor for driving the inner baffle 10 to rotate are further disposed inside the black box 7, wherein the inner baffle 10 is located above the second conveying mechanism 11-2; the second driving motor is arranged on the inner wall of the black box 7 and is connected with the inner baffle 10. The inner baffle 10 is driven to rotate by the second driving motor, so that the function of stopping the movement of the salted duck eggs 8 is achieved. When the salted duck eggs 8 are detected, the second driving motor drives the inner baffle 10 to rotate, so that the inner baffle 10 is driven to block the salted duck eggs 8 and enable the salted duck eggs to be located in the detection range of the detection device all the time, and after the salted duck eggs are detected, the second driving motor drives the inner baffle 10 to rotate, so that the salted duck eggs 8 are released, and the detection of the salted duck eggs 8 is facilitated.
Referring to fig. 1-2, the salted duck egg quality sorting device based on the convolutional neural network further comprises two groups of sorting devices mounted on a rack, wherein the two groups of sorting devices are sequentially arranged along the conveying direction of the third conveying mechanism 11-3, each group of sorting device comprises a conveying claw 16, a driving cylinder 17 for driving the conveying claw 16 to move perpendicular to the conveying direction of the third conveying mechanism 11-3, and a fourth conveying mechanism 18 for conveying the salted duck eggs 8 in the third conveying mechanism 11-3 to the corresponding collecting mechanism 11-4, and the conveying claw 16 is connected with a telescopic rod of the driving cylinder 17; the conveying direction of the fourth conveying mechanism 18 is perpendicular to the conveying direction of the third conveying mechanism 11-3, one end of the fourth conveying mechanism 18 is in butt joint with the third conveying mechanism 11-3, and the other end of the fourth conveying mechanism 18 is in butt joint with the collecting mechanism 11-4. After the salted duck eggs 8 are conveyed to the third conveying mechanism 11-3 by the second conveying mechanism 11-2, the salted duck eggs 8 are conveyed to corresponding sorting devices by the third conveying mechanism 11-3 according to different qualities, such as two sorting devices in the embodiment, wherein one of the two sorting devices is used for conveying the salted duck eggs 8 to the unqualified collecting mechanism 11-4; the salted duck eggs 8 are sent into a qualified collecting mechanism 11-4 by the other sorting device; the corresponding driving air cylinder 17 of the sorting device drives the carrying claw 16 to move, so that the salted duck eggs 8 are driven to move in the direction perpendicular to the conveying direction of the third conveying mechanism 11-3, the salted duck eggs 8 are conveyed to the corresponding fourth conveying mechanism 18, and the salted duck eggs are conveyed to the corresponding collecting mechanism 11-4 through the fourth conveying mechanism 18, so that the sorting work of the salted duck eggs 8 is completed.
In addition to the above embodiments, the sorting device may also adopt a mechanical arm sorting and conveying manner to convey salted duck eggs 8 with different qualities to different collecting mechanisms 11-4, wherein the mechanical arm sorting and conveying manner may be implemented by referring to a sorting device existing in the market, for example, an "egg sorting device" disclosed in the invention patent application with application publication number CN 109349164A.
The first conveying mechanism 11-1, the second conveying mechanism 11-2, the third conveying mechanism 11-3 and the fourth conveying mechanism 18 in the embodiment are synchronous belt transmission mechanisms, and a synchronous belt wheel is driven to rotate through a motor so as to drive a synchronous belt to move, so that the salted duck eggs 8 are conveyed, wherein a limiting block 19 is arranged on the synchronous belt, the limiting block 19 can be a shifting tooth on the synchronous belt and used for preventing the salted duck eggs 8 from moving relative to the synchronous belt in the conveying process, the limiting block 19 can be arranged on the front side and the rear side of the salted duck eggs 8, the cross section of the limiting block 19 is semicircular or circular, and therefore the conveying of the salted duck eggs 8 is facilitated, and the limitation on the relative movement between the salted duck eggs 8 and the synchronous belt is facilitated; the first conveying mechanism 11-1, the second conveying mechanism 11-2, the third conveying mechanism 11-3 and the fourth conveying mechanism 18 can be provided with flanges at the left and right sides of the limiting block 19 for preventing salted duck eggs 8 from rolling down from the synchronous belt.
In addition, the advantage of setting the cross section of the limiting block 19 to be semicircular or circular is that: when the salted duck eggs 8 are conveyed to the detection device by the second conveying mechanism 11-2 to be detected, the second driving motor drives the inner baffle plate 10 to rotate, so that the movement of the salted duck eggs 8 along with the synchronous belt of the second conveying mechanism 11-2 is stopped, and in this way, the salted duck eggs 8 and the second conveying mechanism 11-2 can move relatively, but because the section of the limiting block 19 on the synchronous belt of the second conveying mechanism 11-2 is semicircular or circular, when the salted duck eggs 8 and the limiting block 19 move relatively, the salted duck eggs 8 roll along the outer contour of the limiting block 19, the salted duck eggs 8 can be always limited in the detection range of the detection device, and meanwhile, no damage can be caused to the salted duck eggs 8.
Referring to fig. 1-2, the operation principle of the salted duck egg quality sorting device based on the convolutional neural network is as follows:
during work, the first conveying mechanism 11-1 drives the salted duck eggs 8 to move, the salted duck eggs 8 are conveyed to the second conveying mechanism 11-2, and then the salted duck eggs 8 are conveyed to the detection range of the detection device by the second conveying mechanism 11-2; then, the first driving motor drives the inlet baffle 9 to move, so that the salted duck eggs 8 in the black box 7 are blocked; after the salted duck eggs 8 are detected by the infrared sensor 4, the control system controls the fan 13 to work, the smell of the salted duck eggs 8 is blown to the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5, and the detection information is transmitted to the control system by the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5; if the detected content of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs 8 exceeds a standard value, the salted duck eggs 8 are unqualified; the second conveying mechanism 11-2 conveys the salted duck eggs 8 to a third conveying mechanism 11-3, and the salted duck eggs are conveyed to a collecting mechanism 11-4 for collecting unqualified salted duck eggs through the third conveying mechanism 11-3; if the detected content of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs 8 does not exceed the standard value, the control system controls the image collecting device to work, photographs of the salted duck eggs 8 are taken, and the photograph information is sent to the control system; processing the pictures through a control system, building a convolutional neural network model, and analyzing the quality of the salted duck eggs 8 through the convolutional neural network model; the control system controls the third conveying mechanism 11-3 to convey the detected salted duck eggs 8 to the corresponding collecting mechanism 11-4 according to the analysis result of the convolutional neural network model, so that automation and no humanization of quality sorting work of the salted duck eggs 8 are achieved. And because the data collected in the invention not only comprise the smell of the salted duck egg 8, the cracks of the eggshell of the salted duck egg 8, but also the color, the spots and the like of the eggshell of the salted duck egg 8, the quality analysis of the salted duck egg 8 without cracks can be realized through the data, and the detection precision is higher.
Referring to fig. 1-6, the salted duck egg quality sorting method based on the convolutional neural network comprises the following steps:
(1) firstly, the first conveying mechanism 11-1 drives the salted duck eggs 8 to move, the salted duck eggs 8 are conveyed to the second conveying mechanism 11-2, and then the salted duck eggs 8 are conveyed to the detection range of the detection device by the second conveying mechanism 11-2; the first driving motor drives the inlet baffle 9 to move, so that the salted duck eggs 8 in the black box 7 are blocked from being imported;
(2) the control system controls the fan 13 to work, the odor of the salted duck eggs 8 is blown to the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5, and the detection information is transmitted to the control system by the ammonia gas sensor 3 and the hydrogen sulfide gas sensor 5; if the detected content of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs 8 exceeds a standard value, the salted duck eggs 8 are unqualified; the second conveying mechanism 11-2 conveys the salted duck eggs 8 to a third conveying mechanism 11-3, and the salted duck eggs are conveyed to a collecting mechanism 11-4 for collecting unqualified salted duck eggs 8 through the third conveying mechanism 11-3;
(3) if the detected content of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs 8 does not exceed the standard value, the control system controls the image collecting device to work, shoots the photos of the salted duck eggs 8 and sends the information of the photos to the control system; the control system constructs a convolutional neural network model by processing the picture, wherein the construction of the convolutional neural network model comprises the following steps:
(3-1) sample IMAGE collection and sample library establishment, wherein the collected sample IMAGEs comprise color photographs (namely RGB-IMAGE in the specification of figure 6) of salted duck eggs 8 with different degrees of superiority and inferiority and taken at different angles; carrying out category marking on the obtained photos of the salted duck eggs 8, and specifically comprises the following steps: the salted duck eggs 8 with the eggs of smooth appearance, complete and clean eggshells, no cracks and cyan shells are marked as positive type; marking the salted duck eggs 8 with dark shells, more white or black spots on the surfaces or cracks on the surfaces as the reverse category; if the input pictures contain pictures which do not contain the salted duck eggs 8, marking the pictures as irrelevant pictures; saving 70% of the labeled images as a training sample set and 30% as a testing sample set;
(3-2) normalizing the input samples; normalizing the input samples in the training sample set and the test sample set to 227 multiplied by 227 (pixels) for subsequent input into the convolutional neural network, so as to improve the training efficiency of the convolutional neural network;
(3-3) building a convolutional neural network model; the constructed convolutional neural network model is an AlexNet model, and comprises 5 convolutional layers (CONV), 3 maximum pooling layers (POOL), 3 full-link layers (FC) and 1 output layer; wherein the first convolutional layer uses a convolutional kernel of size 11 × 11 with a step size of 4, followed by a maximum pooling layer of size 3 × 3 (pixels) with a step size of 2 (pixels); the second convolution layer adopts convolution kernels with the size of 5 multiplied by 5, the image size can be compressed by adopting image filling operation to keep the image size unchanged, namely, a circle of blank pixels are filled outside the original image before convolution (the convolution result is not influenced), and the image size after convolution can be kept unchanged; next to a maximum pooling layer of size 3 × 3, the step size is 2; subsequently, the third, fourth and fifth convolutional layers are three tightly connected identical convolutional layers of size 3 × 3, and the image size is kept unchanged by image filling operation; next to a maximum pooling layer of size 3 × 3, the step size is 2; followed by a fully-connected layer with an output of 9216 dimensions, followed by two identical fully-connected layers with an output of 4096 dimensions, followed by the output layer; the output layer adopts a Softmax function as an excitation function, and the rest network layers adopt a Relu function as an excitation function;
(3-4) initializing a convolutional neural network model; initializing by using an Msra method, wherein when only the input number is considered, the Msra initialization is Gaussian distribution with the mean value of 0 and the variance of 2/n;
(3-5) training a convolutional neural network model;
(3-6) classifying the salted duck eggs 8 through a trained convolutional neural network model; classifying the input salted duck egg 8 pictures by a convolutional neural network model, namely classifying the normally shot pictures containing the salted duck egg 8 into a positive type or a negative type; classifying the pictures of the non-salted duck eggs 8 shot by mistake into irrelevant classes;
(4) and the control system controls the third conveying mechanism 11-3 to convey the salted duck eggs 8 after detection to the corresponding collecting mechanism 11-4 according to the classification result of the convolutional neural network model.
Referring to fig. 1 to 6, in step (3-5), the step of training the convolutional neural network model is:
(3-51), inputting characteristic variables;
(3-52) carrying out logic calculation on each item of data in the characteristic variables, and adding a normalization item during calculation;
(3-53) calculating the output of each neuron of each layer, and randomly abandoning the output of a part of neurons in the first two fully-connected layers by adopting a Dropout method;
(3-54) calculating a cost function of the logistic regression;
(3-55) calculating a proper weight matrix and a proper bias vector value by using a gradient descent method, and minimizing a cost function of logistic regression;
(3-56) repeating the steps (3-53) - (3-55) until the prediction accuracy meets the requirement;
and (3-57) analyzing results and outputting the results.
Example 2
Referring to fig. 7, the present embodiment is different from embodiment 1 in that: the conveying speed of the first conveying mechanism 11-1 is less than that of the second conveying mechanism 11-2, namely V2 is greater than V1. Thus, when the salted duck eggs 8 are transferred from the first conveying mechanism 11-1 to the second conveying mechanism 11-2, the conveying speed of the second conveying mechanism 11-2 is greater than that of the first conveying mechanism 11-1, which is equivalent to acceleration of the salted duck eggs 8, and a horizontal forward acting force is generated, so that the salted duck eggs 8 and the subsequent salted duck eggs 8 are pulled apart by a certain distance, for example, when the speed of V2 is twice that of V1, the distance between the salted duck eggs 8 entering the second conveying mechanism 11-2 is twice that of the first conveying mechanism 11-1, so that the inlet baffle 9 can be ensured to smoothly close the inlet of the salted duck eggs without damaging the salted duck eggs 8 by reasonably adjusting the parameters of V1 and V2 according to the detection time; meanwhile, the first driving motor can drive the inlet baffle 9 to rotate, so that the salted duck eggs 8 at the back are smoothly blocked, the salted duck eggs 8 can be detected one by one, and the salted duck eggs 8 can be prevented from being damaged.
In addition to the above-described structure, the remaining structure can be implemented with reference to embodiment 1.
Example 3
Referring to fig. 8, the present embodiment is different from embodiment 1 in that the third conveying mechanisms 11-3 are two groups, an included angle between one group of the third conveying mechanisms 11-3 and the second conveying mechanism 11-2 is 135 degrees, an included angle between the other group of the third conveying mechanisms 11-3 and the second conveying mechanism 11-2 is 225 degrees, and the two groups of the third conveying mechanisms 11-3 are respectively butted with the second conveying mechanism 11-2, wherein a head end of each group of the third conveying mechanisms 11-3 is communicated with a tail end of the second conveying mechanism 11-2, and the tail end is connected with each collecting mechanism 11-4; the tail end of the second conveying mechanism 11-2 is provided with a sorting baffle 14, and the sorting baffle 14 is driven by a third driving motor and is used for conveying salted duck eggs 8 into different third conveying mechanisms 11-3; when the third driving motor drives the sorting baffle plate 14 to rotate to form an included angle of 135 degrees with the second conveying mechanism 11-2, the salted duck eggs 8 coming out of the second conveying mechanism 11-2 are in contact with the sorting baffle plate 14, move to the third conveying mechanism 11-3 forming an included angle of 135 degrees with the second conveying mechanism 11-2 along the inclined direction of the sorting baffle plate 14 under the driving of the second conveying mechanism 11-3, and are conveyed to the corresponding collecting mechanism 11-4 by the third conveying mechanism 11-3; similarly, when the third driving motor drives the sorting baffle 14 to rotate to form an included angle of 225 degrees with the second conveying mechanism 11-2, the salted duck eggs 8 can enter the third conveying mechanism 11-3 which forms an included angle of 225 degrees with the second conveying mechanism 11-2 under the guidance of the sorting baffle 14. Thus, the salted duck eggs 8 with different qualities can be sorted.
In this embodiment, both sides of the synchronous belts of the first conveying mechanism 11-1, the second conveying mechanism 11-2 and the third conveying mechanism 11-3 are provided with flanges, such as the flange 12 at both sides of the first conveying mechanism 11-1 and the flange 15 at both sides of the third conveying mechanism 11-3; this prevents the salted duck eggs 8 from falling down from the timing belt.
In addition to the above-described structure, the remaining structure can be implemented with reference to embodiment 1.
The present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents and are included in the scope of the present invention.
Claims (10)
1. A salted duck egg quality sorting device based on a convolutional neural network is characterized by comprising a rack, a black box arranged on the rack, a detection device arranged in the black box, a conveying device used for conveying salted duck eggs into the black box for detection and a control system, wherein,
two sides of the black box are respectively provided with a salted duck egg inlet and a salted duck egg outlet, wherein the salted duck egg inlet is provided with an inlet baffle and a first driving motor for driving the inlet baffle to rotate so as to enable salted duck eggs to enter the black box one by one, the first driving motor is installed on the black box, and a main shaft of the first driving motor is connected with the inlet baffle;
the conveying device comprises a first conveying mechanism, a second conveying mechanism and a third conveying mechanism, wherein the second conveying mechanism penetrates through the black box, the head end of the second conveying mechanism is connected with the tail end of the first conveying mechanism, and the tail end of the second conveying mechanism is connected with the head end of the third conveying mechanism; the third conveying mechanism is used for conveying the detected salted duck eggs to different collecting mechanisms;
the detection device comprises a first detection module arranged in the black box and a second detection module used for detecting the surface of the salted duck eggs, wherein,
the first detection module comprises an ammonia gas sensor, a hydrogen sulfide gas sensor and a fan which are arranged in a black box, wherein the ammonia gas sensor and the hydrogen sulfide gas sensor are arranged on the inner wall of the black box and are positioned on one side of the second conveying mechanism and used for detecting whether ammonia gas and hydrogen sulfide gas are contained in the gas emitted by salted duck eggs; the fan is arranged on the inner wall of the black box, is arranged opposite to the ammonia gas sensor and the hydrogen sulfide gas sensor, and is used for blowing gas emitted by salted duck eggs into the detection range of the ammonia gas sensor and the hydrogen sulfide gas sensor;
the second detection module comprises an image collection device, and is used for collecting the images of the salted duck eggs detected in the black box, transmitting the images to the control system, and analyzing and processing the images by the control system.
2. The convolutional neural network-based salted duck egg quality sorting device as claimed in claim 1, wherein the image collecting device comprises an industrial camera, the industrial camera is arranged on the top of the inner wall of the black box and is positioned above the second conveying mechanism.
3. The salted duck egg quality sorting device based on the convolutional neural network as claimed in claim 2, wherein the image collecting device further comprises light supplementing lamps, and the light supplementing lamps are installed on two sides of the industrial camera and used for supplementing light.
4. The salted duck egg quality sorting device based on the convolutional neural network as claimed in claim 1, wherein an infrared sensor is arranged between the ammonia gas sensor and the hydrogen sulfide gas sensor for detecting whether salted duck eggs are detected.
5. The salted duck egg quality sorting device based on the convolutional neural network as claimed in claim 1, wherein an inner baffle and a second driving motor for driving the inner baffle to rotate are further arranged inside the black box, and the inner baffle is located above the second conveying mechanism; the second driving motor is installed on the inner wall of the black box, and a main shaft of the second driving motor is connected with the inner baffle.
6. The convolutional neural network-based salted duck egg quality sorting device as claimed in claim 1, wherein the conveying speed of the first conveying mechanism is less than that of the second conveying mechanism.
7. The salted duck egg quality sorting device based on the convolutional neural network as claimed in claim 1, wherein the number of the third conveying mechanisms is multiple, the head end of each group of the third conveying mechanisms is communicated with the tail end of the second conveying mechanism, and the tail ends are connected with the collecting mechanisms; the end of the second conveying mechanism is provided with a sorting baffle, and the sorting baffle is driven by a third driving motor and is used for conveying salted duck eggs into a corresponding third conveying mechanism.
8. The salted duck egg quality sorting device based on the convolutional neural network of claim 1, wherein the first conveying mechanism, the second conveying mechanism and the third conveying mechanism are all synchronous belt transmission mechanisms.
9. A salted duck egg quality sorting method for the convolutional neural network-based salted duck egg quality sorting device according to any one of claims 1 to 8, comprising the steps of:
(1) firstly, the first conveying mechanism drives salted duck eggs to move, the salted duck eggs are conveyed to the second conveying mechanism, and then the salted duck eggs are conveyed to the detection range of the detection device by the second conveying mechanism; then, the first driving motor drives the inlet baffle to move, and the salted duck egg inlet in the black box is blocked;
(2) the control system controls the fan to work, the odor of the salted duck eggs is blown to the ammonia gas sensor and the hydrogen sulfide gas sensor, and the detection information is transmitted to the control system by the ammonia gas sensor and the hydrogen sulfide gas sensor; if the detected contents of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs exceed the standard values, the salted duck eggs are unqualified; the second conveying mechanism conveys the salted duck eggs to a third conveying mechanism, and the salted duck eggs are conveyed to a collecting mechanism for collecting unqualified salted duck eggs through the third conveying mechanism;
(3) if the detected content of ammonia gas and hydrogen sulfide gas emitted by the salted duck eggs does not exceed the standard value, the control system controls the image collecting device to work, photographs of the salted duck eggs are shot, and the photograph information is sent to the control system; the control system constructs a convolutional neural network model by processing the picture, wherein the construction of the convolutional neural network model comprises the following steps:
(3-1) collecting sample images and establishing a sample library, wherein the collected sample images comprise color photographs of salted duck eggs with different advantages and disadvantages taken at different angles; carrying out category marking on the obtained photos of the salted duck eggs, and the method comprises the following steps: the salted duck eggs with round and smooth appearance, complete and clean eggshells, no cracks and cyan shells are marked as positive type; marking salted duck eggs with dark shells, more white or black spots on the surfaces or cracks on the surfaces as reverse categories; if the input pictures contain pictures which do not contain salted duck eggs, marking the pictures as irrelevant pictures; then saving 70% of the labeled images as a training sample set and 30% as a testing sample set;
(3-2) normalizing the input sample image, and normalizing the input sample image in the training sample set and the test sample set to 227 multiplied by 227 for subsequent input into the convolutional neural network;
(3-3) building a convolutional neural network model; the constructed convolutional neural network model is an AlexNet model, and the network layer of the convolutional neural network model comprises 5 convolutional layers, 3 maximum pooling layers, 3 full-connection layers and 1 output layer; wherein, the first convolution layer adopts convolution kernel with the size of 11 multiplied by 11 and the step length is 4, and then a maximum pooling layer with the size of 3 multiplied by 3 is followed and the step length is 2; the second convolution layer adopts convolution kernel with the size of 5 multiplied by 5, the image size is kept unchanged by adopting image filling operation, and then a maximum pooling layer with the size of 3 multiplied by 3 is followed, and the step length is 2; subsequently, the third, fourth and fifth convolutional layers are three tightly connected identical convolutional layers of size 3 × 3, and the image size is kept unchanged by image filling operation; next to a maximum pooling layer of size 3 × 3, the step size is 2; followed by a fully-connected layer with an output of 9216 dimensions, followed by two identical fully-connected layers with an output of 4096 dimensions, followed by the output layer; the output layer adopts a Softmax function as an excitation function, and the rest network layers adopt a Relu function as an excitation function;
(3-4) initializing a convolution neural network model, initializing by using an Msra algorithm, and when only the input number is considered, initializing the Msra to be Gaussian distribution with the mean value of 0 and the variance of 2/n;
(3-5) training a convolutional neural network model;
(3-6) classifying the salted duck eggs through a trained convolutional neural network model; classifying the normally shot pictures containing the salted duck eggs into a positive type or a negative type; classifying the pictures of the non-salted duck eggs shot by mistake into irrelevant classes;
(4) and the control system controls the third conveying mechanism to convey the salted duck eggs after detection to the corresponding collecting mechanism according to the classification result of the convolutional neural network model.
10. The salted duck egg quality sorting method based on the convolutional neural network as claimed in claim 9, wherein in the step (3-5), the step of training the convolutional neural network model is as follows:
(3-51), inputting characteristic variables;
(3-52) carrying out logic calculation on each item of data in the characteristic variables, and adding a normalization item during calculation;
(3-53) calculating the output of each neuron of each layer, and randomly abandoning the output of a part of neurons in the first two fully-connected layers by adopting a Dropout method;
(3-54) calculating a cost function of the logistic regression;
(3-55) calculating a proper weight matrix and a proper bias vector value by using a gradient descent method, and minimizing a cost function of logistic regression;
(3-56) repeating the steps (3-53) - (3-55) until the prediction accuracy meets the requirement;
and (3-57) analyzing and outputting results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010037406.2A CN111136027B (en) | 2020-01-14 | 2020-01-14 | Salted duck egg quality sorting device and method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010037406.2A CN111136027B (en) | 2020-01-14 | 2020-01-14 | Salted duck egg quality sorting device and method based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111136027A true CN111136027A (en) | 2020-05-12 |
| CN111136027B CN111136027B (en) | 2024-04-12 |
Family
ID=70524905
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010037406.2A Active CN111136027B (en) | 2020-01-14 | 2020-01-14 | Salted duck egg quality sorting device and method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111136027B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111790639A (en) * | 2020-05-22 | 2020-10-20 | 北京正大蛋业有限公司 | Poultry egg transportation is with assembly line turnover case letter sorting system |
| CN112136722A (en) * | 2020-08-28 | 2020-12-29 | 江苏理工学院 | Egg sorting system and egg sorting method |
| CN112317338A (en) * | 2020-10-15 | 2021-02-05 | 兰州大学 | Piston ring detection system |
| CN113945560A (en) * | 2021-10-15 | 2022-01-18 | 高邮市双欣蛋品有限公司 | Salted duck egg processing is with automatic light inspection device |
| CN114342838A (en) * | 2022-01-08 | 2022-04-15 | 河北农业大学 | Egg dark spot detection and automatic grading system based on deep learning or image recognition |
| CN114532253A (en) * | 2022-04-26 | 2022-05-27 | 华南农业大学 | Automatic intelligent detection device for hatching egg activity |
| CN114544772A (en) * | 2022-04-26 | 2022-05-27 | 华南农业大学 | Device and method for detecting duck egg cracks based on convolutional neural network and voice frequency spectrum |
| CN117546800A (en) * | 2023-11-17 | 2024-02-13 | 江苏省家禽科学研究所 | Internet of things intelligent detection and distinguishing device and method for black eggs and common eggs |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101419213A (en) * | 2008-12-08 | 2009-04-29 | 浙江大学 | Bird egg crack detection device and method by utilizing volatile matter |
| CN101419212A (en) * | 2008-12-08 | 2009-04-29 | 浙江大学 | Bird egg freshness detection device and method by utilizing volatile matter |
| CN102179374A (en) * | 2010-12-23 | 2011-09-14 | 华中农业大学 | Automatic detecting and sorting device for poultry egg quality and method thereof |
| CN106778902A (en) * | 2017-01-03 | 2017-05-31 | 河北工业大学 | Milk cow individual discrimination method based on depth convolutional neural networks |
| CN108051449A (en) * | 2018-01-30 | 2018-05-18 | 华中农业大学 | The online visible detection method of Salted duck egg face crack based on morphologic edge detection |
| CN108446729A (en) * | 2018-03-13 | 2018-08-24 | 天津工业大学 | Egg embryo classification method based on convolutional neural networks |
| CN108663367A (en) * | 2018-03-30 | 2018-10-16 | 中国农业大学 | A kind of egg quality lossless detection method based on egg unit weight |
| CN108700528A (en) * | 2016-02-17 | 2018-10-23 | 股份公司南备尔 | The surface examining device of egg |
| CN109937912A (en) * | 2019-04-08 | 2019-06-28 | 武汉科技大学 | A kind of egg categorizing system and method based on machine vision |
| CN110006899A (en) * | 2019-04-12 | 2019-07-12 | 华中农业大学 | The lossless vision detection and classification method of lime-preserved egg inside quality |
| CN110455806A (en) * | 2018-05-07 | 2019-11-15 | 南京农业大学 | A kind of egg dynamic image acquisition equipment |
| CN211756981U (en) * | 2020-01-14 | 2020-10-27 | 广东技术师范大学 | Salted duck egg quality sorting device based on convolutional neural network |
-
2020
- 2020-01-14 CN CN202010037406.2A patent/CN111136027B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101419213A (en) * | 2008-12-08 | 2009-04-29 | 浙江大学 | Bird egg crack detection device and method by utilizing volatile matter |
| CN101419212A (en) * | 2008-12-08 | 2009-04-29 | 浙江大学 | Bird egg freshness detection device and method by utilizing volatile matter |
| CN102179374A (en) * | 2010-12-23 | 2011-09-14 | 华中农业大学 | Automatic detecting and sorting device for poultry egg quality and method thereof |
| CN108700528A (en) * | 2016-02-17 | 2018-10-23 | 股份公司南备尔 | The surface examining device of egg |
| CN106778902A (en) * | 2017-01-03 | 2017-05-31 | 河北工业大学 | Milk cow individual discrimination method based on depth convolutional neural networks |
| CN108051449A (en) * | 2018-01-30 | 2018-05-18 | 华中农业大学 | The online visible detection method of Salted duck egg face crack based on morphologic edge detection |
| CN108446729A (en) * | 2018-03-13 | 2018-08-24 | 天津工业大学 | Egg embryo classification method based on convolutional neural networks |
| CN108663367A (en) * | 2018-03-30 | 2018-10-16 | 中国农业大学 | A kind of egg quality lossless detection method based on egg unit weight |
| CN110455806A (en) * | 2018-05-07 | 2019-11-15 | 南京农业大学 | A kind of egg dynamic image acquisition equipment |
| CN109937912A (en) * | 2019-04-08 | 2019-06-28 | 武汉科技大学 | A kind of egg categorizing system and method based on machine vision |
| CN110006899A (en) * | 2019-04-12 | 2019-07-12 | 华中农业大学 | The lossless vision detection and classification method of lime-preserved egg inside quality |
| CN211756981U (en) * | 2020-01-14 | 2020-10-27 | 广东技术师范大学 | Salted duck egg quality sorting device based on convolutional neural network |
Non-Patent Citations (1)
| Title |
|---|
| 雷伏元: "《规则形体物品的包装计量》", vol. 1, 天津科学技术出版社, pages: 96 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111790639A (en) * | 2020-05-22 | 2020-10-20 | 北京正大蛋业有限公司 | Poultry egg transportation is with assembly line turnover case letter sorting system |
| CN111790639B (en) * | 2020-05-22 | 2021-12-10 | 北京正大蛋业有限公司 | Poultry egg transportation is with assembly line turnover case letter sorting system |
| CN112136722A (en) * | 2020-08-28 | 2020-12-29 | 江苏理工学院 | Egg sorting system and egg sorting method |
| CN112317338A (en) * | 2020-10-15 | 2021-02-05 | 兰州大学 | Piston ring detection system |
| CN113945560A (en) * | 2021-10-15 | 2022-01-18 | 高邮市双欣蛋品有限公司 | Salted duck egg processing is with automatic light inspection device |
| CN114342838A (en) * | 2022-01-08 | 2022-04-15 | 河北农业大学 | Egg dark spot detection and automatic grading system based on deep learning or image recognition |
| CN114342838B (en) * | 2022-01-08 | 2022-12-09 | 河北农业大学 | Egg dark spot detection and automatic grading system based on deep learning or image recognition |
| CN114532253A (en) * | 2022-04-26 | 2022-05-27 | 华南农业大学 | Automatic intelligent detection device for hatching egg activity |
| CN114544772A (en) * | 2022-04-26 | 2022-05-27 | 华南农业大学 | Device and method for detecting duck egg cracks based on convolutional neural network and voice frequency spectrum |
| CN114532253B (en) * | 2022-04-26 | 2022-07-22 | 华南农业大学 | Automatic intelligent detection device for hatching egg activity |
| CN117546800A (en) * | 2023-11-17 | 2024-02-13 | 江苏省家禽科学研究所 | Internet of things intelligent detection and distinguishing device and method for black eggs and common eggs |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111136027B (en) | 2024-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111136027A (en) | Salted duck egg quality sorting device and method based on convolutional neural network | |
| CN106238342B (en) | Panoramic vision potato sorts and defect detecting device and its sorting detection method | |
| CN107944504B (en) | Board recognition and machine learning method and device for board recognition and electronic equipment | |
| CN109724984B (en) | Defect detection and identification device and method based on deep learning algorithm | |
| CN100547394C (en) | Fruit quality detection system based on image information fusion technology | |
| CN112452801A (en) | High-efficiency online detection device for defects of glass bottles and working method thereof | |
| CN102179374B (en) | Automatic detecting and sorting device for poultry egg quality and method thereof | |
| JP7129669B2 (en) | Labeled image data creation method, inspection method, program, labeled image data creation device and inspection device | |
| JP7186304B2 (en) | Defect inspection device | |
| CN109187553B (en) | On-line intelligent identification method for deteriorated eggs based on machine vision | |
| CN1485616A (en) | Fowl eggs quality non-destruction automatic detection grading apparatus and process | |
| EP3745298B1 (en) | Classification method and system for high-throughput transparent articles | |
| CN110694921A (en) | Vegetable sorting system and control method thereof | |
| TWI700129B (en) | Moving object classification system and method using self-learning technology | |
| CN113145492A (en) | Visual grading method and grading production line for pear appearance quality | |
| CN211756981U (en) | Salted duck egg quality sorting device based on convolutional neural network | |
| JP2024009979A (en) | Egg classification device, egg classification method, and computer program | |
| CN114354637A (en) | Fruit quality comprehensive grading method and device based on machine vision and X-ray | |
| CN204346935U (en) | The location of food multi-packaging thing nondestructive detection system and dust guard | |
| WO2021221176A1 (en) | Inspection device for tofu products, manufacturing system for tofu products, inspection method for tofu products, and program | |
| CN113077450A (en) | Cherry grading detection method and system based on deep convolutional neural network | |
| CN112950602A (en) | Candy defect detection device and classification method based on depth camera and machine learning | |
| CN117546800A (en) | Internet of things intelligent detection and distinguishing device and method for black eggs and common eggs | |
| CN114677597B (en) | Gear defect visual inspection method and system based on improved YOLOv5 network | |
| CN107742355B (en) | Automatic device of conveying letter sorting coin |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |