CN108470184A - Food materials recognition methods, identification device and household appliance - Google Patents
Food materials recognition methods, identification device and household appliance Download PDFInfo
- Publication number
- CN108470184A CN108470184A CN201810140841.0A CN201810140841A CN108470184A CN 108470184 A CN108470184 A CN 108470184A CN 201810140841 A CN201810140841 A CN 201810140841A CN 108470184 A CN108470184 A CN 108470184A
- Authority
- CN
- China
- Prior art keywords
- food materials
- deep learning
- training
- training sample
- sample set
- 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
- 235000013305 food Nutrition 0.000 title claims abstract description 255
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000012549 training Methods 0.000 claims abstract description 194
- 238000013135 deep learning Methods 0.000 claims abstract description 114
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000013459 approach Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 235000015219 food category Nutrition 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 235000013311 vegetables Nutrition 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of food materials recognition methods, identification device and household appliance, the recognition methods includes that the process of food materials is identified based on deep learning training pattern, and the deep learning training pattern is obtained using following methods：Holostrome training is carried out according to the first food materials training sample set pair deep learning network, obtains the deep learning basic model based on the first food materials training sample set；It is adjusted according to the parameter of the hidden layer in deep learning basic model described in the second food materials training sample set pair, the training deep learning basic model, obtains the deep learning training pattern based on the first food materials training sample set and the second food materials training sample set.With the application of the invention, the training process of the deep learning training pattern by adjusting food materials for identification, improves the training speed and accuracy of identification of training pattern, and then improve food materials recognition accuracy.
Description
Technical field
The invention belongs to household appliance control technology fields, specifically, be related to food materials recognition methods, identification device and
Household appliance.
Background technology
With the fast development of the progress and artificial intelligence of science and technology, deep learning algorithm also more and more applies to
In daily life, for the storage household appliance of food materials, such as refrigerator, as the higher household of kitchen frequency of use
Electric appliance, intelligent development are vital, and the intelligentized most critical issues, in that stored food materials carry out from
Dynamic identification.
The Chinese patent application of Publication No. CN107024073A discloses a kind of multisensor intelligence based on deep learning
Energy controlling method for refrigerator and intelligent refrigerator, and specifically disclose：It obtains and places food materials previous image into refrigerator storeroom and put
The second image after food materials is set, food materials figure to be measured is obtained using visible detection method based on described first image and the second image
Picture is identified above-mentioned food materials image to be measured using the first good convolution neural network model of pre-training, obtains refrigerator storage
Indoor food materials type to be measured.Wherein, the first convolutional neural networks are that pre-training is good, and not unalterable, can make
With re -training in the process, adjusting parameter.
The training of traditional neural network model based on deep learning is needed using a large amount of food materials training sample, flower
Take a large amount of time to carry out the training of neural network holostrome, just can guarantee the accuracy of identification of trained network model.But
It is for food materials, especially to increase food materials newly, the food materials image that can be provided as training sample is less, and sample is trained using newly-increased food materials
The training of this progress network holostrome, the accuracy of identification of the obtained network model newly-increased food materials less to training samples number
It is low, therefore, cause food materials discrimination poor, affects and identify realized household appliance intelligentized control method based on food materials.
Invention content
The purpose of the present invention is to provide a kind of food materials recognition methods and food materials identification devices, by adjusting eating for identification
The training process of the deep learning training pattern of material, improves the training speed and accuracy of identification of training pattern, and then improves
Food materials recognition accuracy.
For achieving the above object, food materials recognition methods provided by the invention is achieved using following technical proposals：
A kind of food materials recognition methods includes the process that food materials are identified based on deep learning training pattern, which is characterized in that the depth
Learning training model is spent to obtain using following methods：
Holostrome training is carried out according to the first food materials training sample set pair deep learning network, obtains and is trained based on first food materials
The deep learning basic model of sample set；If the deep learning network has input layer, dried layer hidden layer and output layer；
It is adjusted according to the parameter of the hidden layer in deep learning basic model described in the second food materials training sample set pair, training
The deep learning basic model is obtained based on the first food materials training sample set and the second food materials training sample set
Deep learning training pattern.
Method as described above, it is described according to hidden in deep learning basic model described in the second food materials training sample set pair
Parameter containing layer is adjusted, and the training deep learning basic model specifically includes：
It is carried out according to the parameter of one layer of hidden layer in deep learning basic model described in the second food materials training sample set pair
Adjustment, the parameter of input layer, output layer and remaining hidden layer remain unchanged, the training deep learning basic model.
The number of plies of method as described above, the hidden layer is no less than 3 layers, according to the second food materials training sample set pair
The parameter of second layer hidden layer in the deep learning basic model is adjusted.
Method as described above, the sample size that the first food materials training sample is concentrated are instructed much larger than second food materials
Practice the sample size in sample set.
Method as described above, the first food materials training sample set or the second food materials training sample set are using following
Process determines：
Acquisition includes the food materials image of known category, is labeled to the food materials image, obtains food materials and marks image；To institute
It states food materials mark image to pre-process, obtains food materials training sample；Multiple food materials training samples are determined as first food
Material training sample set or the second food materials training sample set.
Method as described above is labeled the food materials image, including：The food materials image is marked using edge
Algorithm carries out food materials edge mark；The food materials position in the food materials image is determined according to food materials edge mark.
To realize that aforementioned invention purpose, food materials identification device provided by the invention are achieved using following technical proposals：
A kind of food materials identification device, includes the food materials recognition unit that food materials are identified based on deep learning training pattern, and feature exists
In described device further includes：
Deep learning basic model acquiring unit, for carrying out holostrome according to the first food materials training sample set pair deep learning network
Training obtains the deep learning basic model based on the first food materials training sample set；The deep learning network has defeated
If entering layer, dried layer hidden layer and output layer；
Deep learning training pattern acquiring unit, for according to deep learning basic model described in the second food materials training sample set pair
In the parameter of hidden layer be adjusted, the training deep learning basic model obtains and trains sample based on first food materials
The deep learning training pattern of this collection and the second food materials training sample set.
Device as described above further includes：
Food materials training sample set determination unit, for obtain include known category food materials image, to the food materials image into
Rower is noted, and is obtained food materials and is marked image；Food materials mark image is pre-processed, food materials training sample is obtained；By multiple institutes
It states food materials training sample and is determined as the first food materials training sample set or the second food materials training sample set.
The present invention also provides a kind of household appliance, for storing food materials, the household appliance includes above-mentioned food materials
Identification device.
Preferably, the household appliance is intelligent refrigerator.
Compared with prior art, the advantages and positive effects of the present invention are：Food materials recognition methods and knowledge provided by the invention
Other device carries out holostrome training first with the first food materials training sample set pair deep learning network, obtains the basic mould of deep learning
Type recycles the parameter of hidden layer in the second food materials training sample set pair deep learning basic model to be adjusted, is used for
The deep learning training pattern of food materials identification, since the second food materials training sample set only carries out the parameter of hidden layer in model
Adjustment can be obtained pair by the number of plies for the hidden layer that Rational choice is adjusted based on the second a small amount of food materials training sample
The higher deep learning training pattern of second food materials discrimination, improves food materials recognition accuracy；And due to only to hidden layer or
Person part hidden layer is adjusted, rather than carries out holostrome adjustment, improves model training speed.
After the specific implementation mode of the present invention is read in conjunction with the figure, the other features and advantages of the invention will become more clear
Chu.
Description of the drawings
Fig. 1 is the flow chart based on food materials recognition methods one embodiment of the present invention；
Fig. 2 is the flow chart based on another embodiment of food materials recognition methods of the present invention；
Fig. 3 is the structure diagram based on food materials identification device one embodiment of the present invention.
Specific implementation mode
In order to make the purpose , technical scheme and advantage of the present invention be clearer, below with reference to drawings and examples,
Invention is further described in detail.
Fig. 1 is referred to, which, which show, is based on food materials recognition methods one embodiment flow chart of the present invention, specifically
Identify the artificial intelligence approach of food materials category.This method is based on deep learning network, it is necessary first to establish deep learning training
Model, then identify based on deep learning training pattern the category of food materials to be identified.
As shown in Figure 1, embodiment realizes that food materials are known method for distinguishing and included the following steps：
Step 11：Holostrome training is carried out according to the first food materials training sample set pair deep learning network, obtains and is based on the first food materials
The deep learning basic model of training sample set.
Wherein, the first food materials training sample set is that food materials data sample known to category is formed by sample set, category packet
Include various vegetables, various fruit etc..Deep learning network is various deep learning networks existing in the prior art, the depth
If it includes input layer, dried layer hidden layer and output layer to practise network.Preferably, deep learning network is the mesh based on deep learning
Mark the network of detection algorithm SSD.
And basis of the deep learning basic model as deep learning training pattern, it is deep by the first food materials training sample set pair
It spends learning network and carries out holostrome training acquisition.To ensure the convergence rate and training precision of deep learning training pattern, the first food
Material training sample set is the set for including great amount of samples.Preferably, the first food materials training sample set select in the prior art at
Data in ripe, to include food materials sample known to a large amount of categories food materials sample database.According to training sample set pair depth
The process that learning network carries out holostrome training and obtains deep learning basic model is realized using the prior art, is not made herein in detail
It is thin to illustrate.
Step 12：Deep learning basic model is trained according to the second food materials training sample set, obtains and is instructed based on the first food materials
Practice the deep learning training pattern of sample set and the second food materials training sample set.
Deep learning basic model in step 11 is for food category included in the first food materials training sample set
Accuracy of identification is higher.If there is there are newly-increased food materials, to improve the identification to increasing food materials newly, need using newly-increased food materials as instruction
Practice sample to be again trained model.The set that newly-increased food materials training sample is constituted is defined as the second food materials instruction by the embodiment
Practice sample set, according to the deep learning basic model that the second food materials training sample set training step 11 obtains, to be based on
The deep learning training pattern of first food materials training sample set and the second food materials training sample set.Wherein, the first food materials train sample
The sample size of this concentration is much larger than the sample size that the second food materials training sample is concentrated.Also, trained according to the second food materials
Sample set is adjusted the parameter of the hidden layer in deep learning basic model, rather than carries out the complete of deep learning basic model
Layer parameter adjusts.It is according in the second food materials training sample set pair deep learning basic model as preferred embodiment
The parameter of wherein one layer hidden layer is adjusted, and the input layer of deep learning basic model, output layer and remaining hidden layer
Parameter remains unchanged, and realizes the training to deep learning basic model.
Embodiment more preferably, in conjunction with the characteristics of food materials image, the hidden layer of selected deep learning network
The number of plies be no less than 3 layers；And according to second of hidden layer in the second food materials training sample set pair deep learning basic model
Parameter is finely adjusted.It is by the second food materials training sample set specifically, for the deep learning network with three layers of hidden layer
It is input to input layer as input data, keeps in deep learning basic model that trained input layer is implicit to first layer
Layer weights are constant, the weights of second of hidden layer of first layer hidden layer pair are constant, weights of the third hidden layer to output layer
It is constant, and only second layer hidden layer instructs the weights of third layer hidden layer according to the condition of convergence etc. required by output layer
Practice, realizes the adjustment to second layer hidden layer to the weights of third layer hidden layer.
Due to the second food materials training sample set as newly-increased food materials, it is difficult to a large amount of training sample is obtained, if still adopted
With the method for carrying out holostrome training according to the second food materials training sample set and the first food materials training sample set pair deep learning network,
Parameter training error can be caused big because the second food materials training sample set quantity is very few, Model Identification rate is low.Also, if into
Row holostrome is trained, and training speed is slow.Therefore, which proposes with the second food materials training sample set pair according to the
The trained deep learning basic model of one food materials training sample set is trained, and is only only to the parameter of hidden layer, even
The parameter of one layer of hidden layer adjusts, and the parameter adjustment of a small amount of number of plies is for holostrome parameter adjustment, the training of the second food materials
Sample set quantity is enough, thus, it is possible to obtain to the higher deep learning training pattern of the second food materials discrimination.Also, it instructs
Practice for speed is trained compared to holostrome and is also greatly improved.
Step 13：Food materials are identified based on deep learning training pattern.
Food materials data to be identified are identified in the deep learning training pattern determined based on step 12, to identify
Food materials category.It uses the method for training pattern identification food materials also for the prior art, does not illustrate specifically herein.
Fig. 2 is referred to, the figure shows based on food materials recognition methods one embodiment flow chart of the present invention, specifically
Identify the artificial intelligence approach of food materials category.This method is based on deep learning network, it is necessary first to establish deep learning training
Model, then identify based on deep learning training pattern the category of food materials to be identified.
As shown in Figure 1, embodiment realizes that food materials are known method for distinguishing and included the following steps：
Step 21：Determine the first food materials training sample set.
First food materials training sample set is that food materials data sample known to category is formed by sample set, specifically product
Food materials image known to class, category include various vegetables, various fruit etc..To ensure the convergence rate of deep learning training pattern
And training precision, the first food materials training sample set are the set for including great amount of samples.Preferably, the first food materials training sample set
Select the food materials image in food materials sample database that is ripe, including food materials sample known to a large amount of categories in the prior art.
Also, determine the first food materials training sample set using following methods：
Acquisition includes the food materials image of known category, is labeled to food materials image, obtains food materials and marks image.As preferred
Embodiment is labeled food materials image, including：Food materials edge mark is carried out using edge dimensioning algorithm to food materials image；
The food materials position determined in food materials image is marked according to food materials edge.Food materials side is carried out using edge dimensioning algorithm to food materials image
Edge marks, and food materials can be avoided to be overlapped and influence the accuracy that food materials mark, and then avoids causing because food materials mark inaccuracy
The discrimination of training pattern reduces.The specific method of edge algorithms progress edge mark is used to may be used food materials image existing
Various edge labelling processes in technology are realized, are not illustrated specifically herein.
Then, food materials mark image is pre-processed, including image normalization, gray scale stretching, denoising, brightness enhancing etc. are pre-
Processing obtains food materials training sample.Finally, multiple food materials training samples are determined as the first food materials training sample set.
Step 22：Holostrome training is carried out according to the first food materials training sample set pair deep learning network, obtains and is based on first
The deep learning basic model of food materials training sample set.
Description of the concrete methods of realizing of the step with reference to 1 embodiment step 11 of figure.
Step 23：Determine the second food materials training sample set.
Deep learning basic model in step 22 is for food category included in the first food materials training sample set
Accuracy of identification is higher.If there is there are newly-increased food materials, to improve the identification to increasing food materials newly, need using newly-increased food materials as instruction
Practice sample to be again trained model.The embodiment increases newly by newly-increased food materials training sample, specifically known to category
The set of food materials image construction is defined as the second food materials training sample set.Second food materials training sample set is used as above-mentioned first eats
The identical method of material training sample set determines that specific method determines the process of the first food materials training sample set referring to step 21.
Step 24：Deep learning basic model is trained according to the second food materials training sample set, obtains and is instructed based on the first food materials
Practice the deep learning training pattern of sample set and the second food materials training sample set.
Description of the specific implementation process of the step with reference to 1 embodiment step 12 of figure.
Step 25：Food materials are identified based on deep learning training pattern.
Food materials data to be identified are identified in the deep learning training pattern determined based on step 24, to identify
Food materials category.It uses the method for training pattern identification food materials also for the prior art, does not illustrate specifically herein.
Fig. 3 is referred to, which show the structure diagram based on food materials identification device one embodiment of the present invention, specific next
It says, is the artificial intelligence device for identifying food materials category.The identification device includes deep learning basic model acquiring unit 31
With deep learning training pattern acquiring unit 32.In some other preferred embodiment, identification device can also include food materials
Training sample set determination unit 33.
Wherein, deep learning basic model acquiring unit 31, for according to the first food materials training sample set pair deep learning
Network carry out holostrome training, obtain based on the first food materials training sample set deep learning basic model.Also, deep learning
If network has input layer, dried layer hidden layer and output layer.
Deep learning training pattern acquiring unit 32, for according to the second basic mould of food materials training sample set pair deep learning
The parameter of the hidden layer in deep learning basic model acquired in type acquiring unit 31 is adjusted, and training deep learning is basic
Model obtains the deep learning training pattern based on the first food materials training sample set and the second food materials training sample set.
Food materials sample set determination unit 33, for obtain include known category food materials image, food materials image is carried out
Mark obtains food materials and marks image；Food materials mark image is pre-processed, food materials training sample is obtained；Multiple food materials are trained
Sample is determined as the first food materials training sample set or the second food materials training sample set.
Above-mentioned each structural unit runs corresponding software program, realizes the recognition methods of Fig. 1 and Fig. 2, realizes to food materials
Identification.The more specific course of work of identification device with above-mentioned each structural unit and the technique effect of generation, with reference to figure 1 and
Fig. 2 shows embodiment description.
Food materials identification device with above-mentioned each structural unit can be applied to the household appliance of storage food materials, such as intelligent ice
In case, accurate, the quick identification to stored food materials is realized.
The above embodiments are merely illustrative of the technical solutions of the present invention, rather than is limited；Although with reference to aforementioned reality
Applying example, invention is explained in detail, for those of ordinary skill in the art, still can be to aforementioned implementation
Technical solution recorded in example is modified or equivalent replacement of some of the technical features；And these are changed or replace
It changes, the spirit and scope for claimed technical solution of the invention that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of food materials recognition methods includes the process for identifying food materials based on deep learning training pattern, which is characterized in that described
Deep learning training pattern is obtained using following methods：
Holostrome training is carried out according to the first food materials training sample set pair deep learning network, obtains and is trained based on first food materials
The deep learning basic model of sample set；If the deep learning network has input layer, dried layer hidden layer and output layer；
It is adjusted according to the parameter of the hidden layer in deep learning basic model described in the second food materials training sample set pair, training
The deep learning basic model is obtained based on the first food materials training sample set and the second food materials training sample set
Deep learning training pattern.
2. recognition methods according to claim 1, which is characterized in that described according to described in the second food materials training sample set pair
The parameter of hidden layer in deep learning basic model is adjusted, and the training deep learning basic model specifically includes：
It is carried out according to the parameter of one layer of hidden layer in deep learning basic model described in the second food materials training sample set pair
Adjustment, the parameter of input layer, output layer and remaining hidden layer remain unchanged, the training deep learning basic model.
3. recognition methods according to claim 2, which is characterized in that the number of plies of the hidden layer is no less than 3 layers, according to institute
The parameter for stating the second layer hidden layer in deep learning basic model described in the second food materials training sample set pair is adjusted.
4. recognition methods according to claim 1, which is characterized in that the sample number that the first food materials training sample is concentrated
Amount is much larger than the sample size that the second food materials training sample is concentrated.
5. recognition methods according to any one of claim 1 to 4, which is characterized in that the first food materials training sample
Collection or the second food materials training sample set are determined using following processes：
Acquisition includes the food materials image of known category, is labeled to the food materials image, obtains food materials and marks image；To institute
It states food materials mark image to pre-process, obtains food materials training sample；Multiple food materials training samples are determined as first food
Material training sample set or the second food materials training sample set.
6. recognition methods according to claim 5, which is characterized in that the food materials image is labeled, including：To institute
It states food materials image and food materials edge mark is carried out using edge dimensioning algorithm；The food materials figure is determined according to food materials edge mark
Food materials position as in.
7. a kind of food materials identification device includes the food materials recognition unit for identifying food materials based on deep learning training pattern, feature
It is, described device further includes：
Deep learning basic model acquiring unit, for carrying out holostrome according to the first food materials training sample set pair deep learning network
Training obtains the deep learning basic model based on the first food materials training sample set；The deep learning network has defeated
If entering layer, dried layer hidden layer and output layer；
Deep learning training pattern acquiring unit, for according to deep learning basic model described in the second food materials training sample set pair
In the parameter of hidden layer be adjusted, the training deep learning basic model obtains and trains sample based on first food materials
The deep learning training pattern of this collection and the second food materials training sample set.
8. identification device according to claim 7, which is characterized in that described device further includes：
Food materials training sample set determination unit, for obtain include known category food materials image, to the food materials image into
Rower is noted, and is obtained food materials and is marked image；Food materials mark image is pre-processed, food materials training sample is obtained；By multiple institutes
It states food materials training sample and is determined as the first food materials training sample set or the second food materials training sample set.
9. a kind of household appliance, for storing food materials, which is characterized in that the household appliance includes the claims 7 or 8
The food materials identification device.
10. household appliance according to claim 9, which is characterized in that the household appliance is intelligent refrigerator.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810140841.0A CN108470184A (en) | 2018-02-11 | 2018-02-11 | Food materials recognition methods, identification device and household appliance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810140841.0A CN108470184A (en) | 2018-02-11 | 2018-02-11 | Food materials recognition methods, identification device and household appliance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108470184A true CN108470184A (en) | 2018-08-31 |
Family
ID=63266514
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810140841.0A Pending CN108470184A (en) | 2018-02-11 | 2018-02-11 | Food materials recognition methods, identification device and household appliance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108470184A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111353333A (en) * | 2018-12-21 | 2020-06-30 | 九阳股份有限公司 | Food material identification method, household appliance and food material identification system |
| CN111523620A (en) * | 2020-07-03 | 2020-08-11 | 北京每日优鲜电子商务有限公司 | Dynamic adjustment method and commodity verification method for commodity identification model |
| CN111860073A (en) * | 2019-04-30 | 2020-10-30 | 青岛海尔智能技术研发有限公司 | Food material image recognition method and device and household appliance |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105512635A (en) * | 2015-12-15 | 2016-04-20 | 鲁东大学 | Category attribute fused deep network underground target identification method and system |
| CN106227851A (en) * | 2016-07-29 | 2016-12-14 | 汤平 | Based on the image search method searched for by depth of seam division that degree of depth convolutional neural networks is end-to-end |
| CN106529564A (en) * | 2016-09-26 | 2017-03-22 | 浙江工业大学 | Food image automatic classification method based on convolutional neural networks |
| WO2017113232A1 (en) * | 2015-12-30 | 2017-07-06 | 中国科学院深圳先进技术研究院 | Product classification method and apparatus based on deep learning |
| CN107358257A (en) * | 2017-07-07 | 2017-11-17 | 华南理工大学 | Under a kind of big data scene can incremental learning image classification training method |
| CN107679552A (en) * | 2017-09-11 | 2018-02-09 | 北京飞搜科技有限公司 | A kind of scene classification method and system based on multiple-limb training |
-
2018
- 2018-02-11 CN CN201810140841.0A patent/CN108470184A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105512635A (en) * | 2015-12-15 | 2016-04-20 | 鲁东大学 | Category attribute fused deep network underground target identification method and system |
| WO2017113232A1 (en) * | 2015-12-30 | 2017-07-06 | 中国科学院深圳先进技术研究院 | Product classification method and apparatus based on deep learning |
| CN106227851A (en) * | 2016-07-29 | 2016-12-14 | 汤平 | Based on the image search method searched for by depth of seam division that degree of depth convolutional neural networks is end-to-end |
| CN106529564A (en) * | 2016-09-26 | 2017-03-22 | 浙江工业大学 | Food image automatic classification method based on convolutional neural networks |
| CN107358257A (en) * | 2017-07-07 | 2017-11-17 | 华南理工大学 | Under a kind of big data scene can incremental learning image classification training method |
| CN107679552A (en) * | 2017-09-11 | 2018-02-09 | 北京飞搜科技有限公司 | A kind of scene classification method and system based on multiple-limb training |
Non-Patent Citations (1)
| Title |
|---|
| 顾月清 等主编: "《全国普通高等院校生物医学工程规划教材 生物医学工程技术》", 北京：中国医药科技出版社 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111353333A (en) * | 2018-12-21 | 2020-06-30 | 九阳股份有限公司 | Food material identification method, household appliance and food material identification system |
| CN111353333B (en) * | 2018-12-21 | 2023-10-20 | 九阳股份有限公司 | Food material identification method, household electrical appliance and food material identification system |
| CN111860073A (en) * | 2019-04-30 | 2020-10-30 | 青岛海尔智能技术研发有限公司 | Food material image recognition method and device and household appliance |
| CN111523620A (en) * | 2020-07-03 | 2020-08-11 | 北京每日优鲜电子商务有限公司 | Dynamic adjustment method and commodity verification method for commodity identification model |
| CN111523620B (en) * | 2020-07-03 | 2020-10-20 | 北京每日优鲜电子商务有限公司 | Dynamic adjustment method and commodity verification method for commodity identification model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109559298A (en) | Emulsion pump defect detection method based on deep learning | |
| CN110310259B (en) | Improved YOLOv3 algorithm-based knot defect detection method | |
| CN108021947B (en) | A kind of layering extreme learning machine target identification method of view-based access control model | |
| CN106355192B (en) | Support vector machine method based on chaos grey wolf optimization | |
| WO2022253057A1 (en) | Intelligent moisture precise irrigation control system and method for fruit and vegetable cultivation in solar greenhouse | |
| CN108470184A (en) | Food materials recognition methods, identification device and household appliance | |
| CN107657603A (en) | A kind of industrial appearance detecting method based on intelligent vision | |
| CN107390754B (en) | Intelligent plant growth environment adjustment system and method based on Internet of Things cloud platform | |
| CN108009526A (en) | A kind of vehicle identification and detection method based on convolutional neural networks | |
| CN109191455A (en) | A kind of field crop pest and disease disasters detection method based on SSD convolutional network | |
| CN112418130A (en) | Banana maturity detection method and device based on BP neural network | |
| CN107390753B (en) | Intelligent plant growth environment regulating device and method based on Internet of Things cloud platform | |
| CN109002769A (en) | A kind of ox face alignment schemes and system based on deep neural network | |
| CN107832780A (en) | Low confidence sample processing method and system are sorted based on artificial intelligence plank | |
| Wang et al. | Towards cooperation in sequential prisoner's dilemmas: a deep multiagent reinforcement learning approach | |
| CN109242830A (en) | A kind of machine vision technique detection method based on deep learning | |
| CN107590799A (en) | The recognition methods of banana maturity period and device based on depth convolutional neural networks | |
| CN112949517B (en) | Plant stomata density and opening degree identification method and system based on deep migration learning | |
| CN110378876A (en) | Image recognition method, device, equipment and storage medium based on deep learning | |
| CN110188780A (en) | Method and device for constructing deep learning model for positioning multi-target feature points | |
| CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
| CN110391022A (en) | A kind of deep learning breast cancer pathological image subdivision diagnostic method based on multistage migration | |
| CN111368900A (en) | Image target object identification method | |
| CN110222738A (en) | Towards mixing sample man sparetime university data based on multiple view dictionary learning classification method | |
| CN116543282A (en) | Crop disease segmentation method, system, equipment and medium based on multi-scale fusion and CBAM-ResNet50 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180831 |
|
| RJ01 | Rejection of invention patent application after publication |