CN107784358A - A kind of food security image detecting method based on LSTM neutral nets - Google Patents

A kind of food security image detecting method based on LSTM neutral nets Download PDF

Info

Publication number
CN107784358A
CN107784358A CN201610721870.7A CN201610721870A CN107784358A CN 107784358 A CN107784358 A CN 107784358A CN 201610721870 A CN201610721870 A CN 201610721870A CN 107784358 A CN107784358 A CN 107784358A
Authority
CN
China
Prior art keywords
lstm
neutral nets
input
food
binary data
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
Application number
CN201610721870.7A
Other languages
Chinese (zh)
Inventor
张业平
张小虎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Suzhou Chuangxin Universal Chromatograph Instrument Co Ltd
Original Assignee
Suzhou Chuangxin Universal Chromatograph Instrument Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Suzhou Chuangxin Universal Chromatograph Instrument Co Ltd filed Critical Suzhou Chuangxin Universal Chromatograph Instrument Co Ltd
Priority to CN201610721870.7A priority Critical patent/CN107784358A/en
Publication of CN107784358A publication Critical patent/CN107784358A/en
Pending legal-status Critical Current

Links

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
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Neurology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The present invention carries out the image recognition of food using LSTM neural network algorithms.LSTM networks have been demonstrated more more efficient than traditional RNNs.In this model, conventional neuron, i.e., one unit that the activation of S types is applied to the combination of its input linear, is replaced by memory cell.Each memory cell is associated with an input gate, an out gate and the noiseless internal state of itself of being sent into of a leap time step.The present invention uses convolutional neural networks（1）Image is switched to the triple channel binary data of rgb format（2）Binary data is input in LSTM neutral nets, calculation formula is as follows

Description

A kind of food security image detecting method based on LSTM neutral nets
Technical field
The present invention relates to food security image detecting technique.
Background technology
In recent years, it is qualified detect whether to need special equipment such as liquid chromatograph for substantial amounts of food, but these set Standby all directly to detect sample, each sample is required for pre-treatment, and flow very bothers.The detection of liquid chromatogram simultaneously Speed is limited, and for substantial amounts of food, liquid chromatographic detection is a hard work, but is necessary.Image is known The development of other art can be that food security opens up a new method.It can use the method for image recognition will be unqualified Food branch away.Use feedforward convolutional neural networks（convnets）Solve computer vision problem with RNNs, be depth Learn the achievement being most widely known by the people, but former information can not be retained for the convolutional neural networks that feedover and feedback, because this is former Cause, cause accuracy rate not high.
The content of the invention
The image recognition of food is carried out using LSTM neural network algorithms for the above situation present invention.LSTM networks are It is proved to more more efficient than traditional RNNs.In this model, conventional neuron, i.e., the activation of S types is applied to it by one The unit of input linear combination, is replaced by memory cell.Each memory cell is and an input gate, an out gate and one It is individual to be associated across the noiseless internal state of itself of being sent into of time step.
The present invention uses convolutional neural networks
（1）Image is switched to the triple channel binary data of rgb format
（2）Binary data is input in LSTM neutral nets, calculation formula is as follows
（3）LSTM neutral nets are automatically classified
It is an advantage of the invention that improve the accuracy rate of food safety detection
Embodiment:
In the model, obtained for each memory cell, three sets of weights from input training, including in previous time step completely Hidden state.One is fed to input node, in the bottom of above formula.One is fed to input gate, at the cell bottom of the rightmost side Portion is shown.Another is fed to out gate, the display in the top rightmost side.Each blue node is related to an activation primitive Connection, typical case is S type functions, and represents the Pi nodes of multiplication.Most central node is referred to as internal state in unit, and Time step is crossed over 1 weight, feeds back to itself.The connection side certainly of internal state, is referred to as constant error conveyer belt or CEC
（1）Image is switched to the triple channel binary data of rgb format
（2）Binary data is input in LSTM neutral nets, calculation formula is as follows
（3）LSTM neutral nets first are trained using a part of data as training data, are carried out automatically with LSTM neutral nets afterwards Classification.

Claims (1)

1. claims：A kind of food security image detecting method based on LSTM neutral nets, it is characterised in that：Using LSTM neural network algorithms carry out the image recognition of food.
CN201610721870.7A 2016-08-25 2016-08-25 A kind of food security image detecting method based on LSTM neutral nets Pending CN107784358A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610721870.7A CN107784358A (en) 2016-08-25 2016-08-25 A kind of food security image detecting method based on LSTM neutral nets

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610721870.7A CN107784358A (en) 2016-08-25 2016-08-25 A kind of food security image detecting method based on LSTM neutral nets

Publications (1)

Publication Number Publication Date
CN107784358A true CN107784358A (en) 2018-03-09

Family

ID=61389126

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610721870.7A Pending CN107784358A (en) 2016-08-25 2016-08-25 A kind of food security image detecting method based on LSTM neutral nets

Country Status (1)

Country Link
CN (1) CN107784358A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111686949A (en) * 2020-07-16 2020-09-22 苏州创新通用色谱仪器有限公司 High-speed vacuum positioning centrifuge
CN114880951A (en) * 2022-06-06 2022-08-09 浙江理工大学 Fabric flaw prediction method based on digital twinning

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105550699A (en) * 2015-12-08 2016-05-04 北京工业大学 CNN-based video identification and classification method through time-space significant information fusion

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105550699A (en) * 2015-12-08 2016-05-04 北京工业大学 CNN-based video identification and classification method through time-space significant information fusion

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
KELVIN XU等: ""Show, Attend and Tell: Neural Image Caption"", 《ARXIV》 *
WONMIN BYEON等: ""Scene analysis by mid-level attribute learning using 2D LSTM networks"", 《PATTERN RECOGNITION LETTERS63》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111686949A (en) * 2020-07-16 2020-09-22 苏州创新通用色谱仪器有限公司 High-speed vacuum positioning centrifuge
CN114880951A (en) * 2022-06-06 2022-08-09 浙江理工大学 Fabric flaw prediction method based on digital twinning

Similar Documents

Publication Publication Date Title
JP2019096006A5 (en)
CN106897714B (en) Video motion detection method based on convolutional neural network
SG11201906576WA (en) Speech wakeup method, apparatus, and electronic device
CN110349103A (en) It is a kind of based on deep neural network and jump connection without clean label image denoising method
CN105678248B (en) Face key point alignment algorithm based on deep learning
CN105550699A (en) CNN-based video identification and classification method through time-space significant information fusion
CN107203891A (en) A kind of automatic many threshold values characteristic filter method and devices
CN104956599A (en) High speed receivers circuits and methods
Yuan et al. Variable step-size sign natural gradient algorithm for sequential blind source separation
CN104573048B (en) A kind of user base attribute forecast method based on smart mobile phone data on flows
CN105022754A (en) Social network based object classification method and apparatus
CN107784358A (en) A kind of food security image detecting method based on LSTM neutral nets
CN107886512A (en) A kind of method for determining training sample
CN108564166A (en) Based on the semi-supervised feature learning method of the convolutional neural networks with symmetrical parallel link
CN110046941A (en) A kind of face identification method, system and electronic equipment and storage medium
Jia et al. Applying data-driven method to production decline analysis and forecasting
CN105512114B (en) The screening technique and system of parallel sentence pairs
Sarussi et al. Towards understanding learning in neural networks with linear teachers
CN106453404B (en) A kind of network inbreak detection method and device
CN104915671A (en) FGAK (Fast Global Alignment Kernels) based abnormal trajectory detection method
CN109614943A (en) A kind of feature extracting method for blind source separating
Chen et al. Topology identification of complex networks from noisy time series using ROC curve analysis
WO2015161293A3 (en) Transaction routing based on customer preference and social contact management
CN106980898A (en) Deep learning system and its application method
CN108197274B (en) Abnormal personality detection method and device based on conversation

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
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20180309