CN114687012A - Efficient foreign fiber removing device and method for high-impurity-content raw cotton - Google Patents

Efficient foreign fiber removing device and method for high-impurity-content raw cotton Download PDF

Info

Publication number
CN114687012A
CN114687012A CN202210176142.8A CN202210176142A CN114687012A CN 114687012 A CN114687012 A CN 114687012A CN 202210176142 A CN202210176142 A CN 202210176142A CN 114687012 A CN114687012 A CN 114687012A
Authority
CN
China
Prior art keywords
cotton
foreign
flow channel
storage box
raw
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
CN202210176142.8A
Other languages
Chinese (zh)
Inventor
罗男
吴泽平
周强兵
魏巍
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.)
Wuhan Zhimu Intelligent Technology Partnership LP
Original Assignee
Wuhan Zhimu Intelligent Technology Partnership LP
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 Wuhan Zhimu Intelligent Technology Partnership LP filed Critical Wuhan Zhimu Intelligent Technology Partnership LP
Priority to CN202210176142.8A priority Critical patent/CN114687012A/en
Publication of CN114687012A publication Critical patent/CN114687012A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • D—TEXTILES; PAPER
    • D01—NATURAL OR MAN-MADE THREADS OR FIBRES; SPINNING
    • D01G—PRELIMINARY TREATMENT OF FIBRES, e.g. FOR SPINNING
    • D01G9/00—Opening or cleaning fibres, e.g. scutching cotton
    • D01G9/04—Opening or cleaning fibres, e.g. scutching cotton by means of beater arms
    • D—TEXTILES; PAPER
    • D01—NATURAL OR MAN-MADE THREADS OR FIBRES; SPINNING
    • D01G—PRELIMINARY TREATMENT OF FIBRES, e.g. FOR SPINNING
    • D01G9/00—Opening or cleaning fibres, e.g. scutching cotton
    • D01G9/14—Details of machines or apparatus
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G06T7/001—Industrial image inspection using an image reference approach
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G06T7/74—Determining position or orientation of objects or cameras using feature-based methods involving reference images or patches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30124—Fabrics; Textile; Paper

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Textile Engineering (AREA)
  • Quality & Reliability (AREA)
  • Preliminary Treatment Of Fibers (AREA)

Abstract

The invention discloses a high-efficiency foreign fiber removing device and a removing method for high-impurity-content raw cotton, relates to the field of cotton foreign fiber clearing, and aims at solving the problem that waste is caused because a great amount of cotton is carried out by removed foreign fibers when an existing foreign fiber machine works, the content of useful cotton fibers is far higher than that of the foreign fibers. The invention does not damage cotton fibers in the process of detecting and removing the foreign fibers, effectively improves the production efficiency of cotton spinning, reduces the labor intensity, reduces the production cost and reduces the waste of raw materials.

Description

Efficient foreign fiber removing device and method for high-impurity-content raw cotton
Technical Field
The invention relates to the field of cotton foreign fiber clearing, in particular to a high-efficiency foreign fiber clearing device and method for raw cotton with high impurity content.
Background
In the production process of the cotton spinning industry, foreign fibers mixed in raw cotton can cause problems of yarn breakage, cloth defects, uneven dyeing and the like, and the product quality is seriously influenced. Therefore, cotton foreign fiber removing machines are increasingly applied to the production flow of the front spinning to remove foreign fibers in raw cotton.
The working principle of the foreign fiber machine is that the whole machine is embedded into a production line, cotton which is fully loosened is guided in through a cotton inlet pipeline, detection and impurity removal are carried out in the foreign fiber machine, and clean cotton is sent to the subsequent working procedures through an outlet pipeline.
The different fiber machine is limited by the working principle and the production line environment, and when the different fiber machine works, a great amount of cotton can be carried out by the removed different fibers. These cotton impurities are raw cotton with high impurity content, and the fiber content of useful cotton is far higher than that of foreign fibers. According to experimental statistics, the weight average proportion of the foreign fibers in the hybrid cotton is only about 0.6%.
The cotton mill generally treats the raw cotton with high impurity content as follows: directly pouring the mixture into a bale plucker for reuse, manually picking the mixture for reuse and selling the mixture to other manufacturers at low price. These treatments all have their inherent disadvantages. The foreign fibers are always in the production line when the foreign fibers are directly poured into the bale plucker for reuse, and are gradually shredded and blended into the yarns, so that the product quality is influenced. The labor intensity of manual picking is high, the working environment is severe, the efficiency is low, and the production cost is increased. Other cotton mills purchasing such cotton waste can only produce low quality products from such raw materials.
In order to solve the problems, a high-efficiency foreign fiber removing device and a removing method for the high-impurity-content raw cotton are provided.
Disclosure of Invention
The invention aims to solve the defects in the prior art, and provides a high-efficiency foreign fiber removing device and a removing method for raw cotton with high impurity content.
In order to achieve the purpose, the invention adopts the following technical scheme:
the utility model provides a high-efficient different fine clearing device to high miscellaneous volume raw cotton, includes opens the device, the top import department of opening the device installs the front end and stores up the hopper, the front end stores up the hopper and opens and be provided with cotton conveyor between the device, the top that the hopper was stored up to the front end is provided with loading attachment, the exit of opening the device is provided with cotton flow channel, cotton flow channel is close to the one end bilateral symmetry of opening the device and is provided with the shooting subassembly, install on the cotton flow channel of shooting subassembly one side and correspond the miscellaneous cotton collection subassembly and the spray valve subassembly that set up in cotton flow channel both sides, the end of miscellaneous cotton collection subassembly is installed and is collected the hopper, the end of cotton flow channel is installed the negative-pressure air fan subassembly through the pipe connection, the one end that cotton flow channel is close to the negative-pressure air fan subassembly has end to store up the hopper through the pipe connection.
Preferably, a motor-driven beating assembly and a plucking roller are arranged in the plucking device.
Preferably, a control component is installed on one side of the tail end cotton storage box.
Preferably, the control assembly comprises an electric cabinet and a display.
Preferably, the photographing assembly includes a light source and an imaging sensing device.
Preferably, the feeding device is one of a feeding hopper, a feeding bag or a crawler feeding.
A high-efficiency foreign fiber removing method for raw cotton with high impurity content comprises the following steps:
the method comprises the following steps: manually pouring the impure cotton beaten out by the foreign fiber machine into a feeding device and storing the impure cotton in a front-end cotton storage box;
step two: conveying the impure cotton in the front end cotton storage box to an opening device by a cotton conveying device, opening the impure cotton by a beater assembly and an opening roller in the opening device, and allowing the opened impure cotton to enter a cotton flow channel;
step three: the cotton flow channel generates negative pressure airflow under the action of the negative pressure fan component to drive the miscellaneous cotton to slowly move towards the tail end cotton storage box in the cotton flow channel;
step four: detecting the impurity cotton by an image detection algorithm when the impurity cotton passes through the shooting assembly;
step five: detecting the position of impurities in the cotton waste by using an image detection algorithm, and hitting the impurities to a cotton waste collection assembly when the cotton waste passes through the eruption assembly;
step six: the cotton flow with the impurities removed enters a tail end cotton storage box for temporary storage, and then the subsequent processing is waited.
Preferably, the image detection algorithm specifically includes: shooting cotton flow images by a shooting component, and inputting two adjacent frames of images; extracting a feature map between two frames of images through a feature extraction network and performing feature matching through a matching network; amplifying the compressed characteristic diagram to the input resolution of an original image by using an up-sampling network; the training is constrained using the pre-labeled position information of both.
Preferably, the constraint training is modified to be a rectangular box area overlapping loss on the basis of a traditional target detection loss function, where the traditional target detection loss function is defined as:
Figure BDA0003520335140000031
wherein B represents the predicted number of target rectangular frames,
Figure BDA0003520335140000032
representing whether the predefined target of the image is contained currently, if the predefined target of the image is contained, the predefined target is 1, otherwise, the predefined target of the image is 0. In addition xi，yi，wi，hiIndicating the predicted coordinate information of the current target
Figure BDA0003520335140000033
The true coordinate information is predefined for the image.
The specific calculation method for the area overlapping loss of the rectangular frame is as follows:
a. defining truth coordinate information as
Figure BDA0003520335140000041
Respectively generationThe initial abscissa, the ordinate, the width and the height of the rectangular frame of the table rectangular frame; this prediction of coordinate information xi，yi，wi，hi；
b. For the pixel point of each position in the predicted rectangular frame, respectively calculating coordinate area information:
X＝(xi+xi+wi)*(yi+yi+hi)
Figure BDA0003520335140000042
Figure BDA0003520335140000043
Figure BDA0003520335140000044
I＝Ih*Iw
Figure BDA0003520335140000045
Figure BDA0003520335140000046
c. from the area information of the coordinates, an overlap area loss function is defined:
Figure BDA0003520335140000047
d. training a multi-scale different fiber detection network by using a new loss function;
e. and stopping training when the accuracy of the network training reaches more than 90% in the verification machine, and storing the model for deployment of the reasoning end.
The invention has the beneficial effects that:
1. the cotton is not damaged in the opening process of the opening device, and the quality of cotton fibers is kept; opening finely, fully separating cotton from impurities, wherein the size of the opened cotton ball can reach 0.01 g; the foreign fiber is not damaged, and the foreign fiber is prevented from being mixed into cotton after being shredded;
2. the cotton flow channel has small thickness and thin cotton layer, so that the foreign fibers are fully exposed, the detection is facilitated, the cost of the whole machine is low, and the occupied area is small;
3. the device can recycle the impure cotton mistakenly hit by the different fiber machine in the previous process, so that the hitting proportion of the different fiber machine is properly increased in a cotton mill, and the product quality is improved;
4. the device runs independently, is not in a fore-spinning production line, has no working procedures and machinery for restricting and influencing cotton flow, has stable cotton flow in a channel, and is simple to install and debug the whole machine;
5. the cotton flow is stable, the requirements on a camera and an electronic board card are low, the imaging sensing device and the spray valve assembly can be matched with a foreign fiber removing algorithm to remove impurities at high speed and accurately, the removing accuracy is improved by 30 times compared with that of a traditional foreign fiber removing machine, and the loss is reduced;
6. the spray valve adopts a high-frequency electromagnetic valve, and can strike 10 ten thousand times per hour; densely arranging 50 valves with the highest density of 50 centimeters; the hitting precision is high, the taken cotton is very little, the foreign fiber removing efficiency is over 95 percent, and the experimental test shows that the foreign fiber can account for over 20 percent of the total weight of sundries hit by the blast valve;
7. the negative pressure fan can adjust the cotton flow speed within a certain range according to the production yield, the cotton quality requirement and the striking efficiency.
Drawings
FIG. 1 is a schematic structural diagram of a foreign fiber removing apparatus according to the present invention;
FIG. 2 is a block diagram of a constraint training network according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
Referring to fig. 1, a high-efficiency foreign fiber removing device for raw cotton with high content of foreign matters comprises an opening device 4, a flapping component and an opening roller driven by a motor are arranged in the opening device 4, a front end cotton storage tank 2 is arranged at an inlet above the opening device 4, a cotton conveying device 3 is arranged between the front end cotton storage tank 2 and the opening device 4, a feeding device 1 is arranged at the top of the front end cotton storage tank 2, the feeding device 1 is one of a feeding hopper, a feeding bag or a crawler feeding, a cotton flow channel 5 is arranged at an outlet of the opening device 4, shooting components 6 are symmetrically arranged at two sides of one end, close to the opening device 4, of the cotton flow channel 5, each shooting component 6 comprises a light source and an imaging sensing device, a foreign cotton collecting component 8 and a spraying valve component 7 which are correspondingly arranged at two sides of the cotton flow channel 5 are arranged on the cotton flow channel 5 at one side of the shooting components 6, a cotton collecting box 12 is arranged at the tail end of the foreign cotton collecting component 8, the end of the cotton flow channel 5 is connected with a negative pressure fan assembly 9 through a pipeline, one end of the cotton flow channel 5 close to the negative pressure fan assembly 9 is connected with an end cotton storage box 10 through a pipeline, a control assembly 11 is installed on one side of the end cotton storage box 10, and the control assembly 11 comprises an electric cabinet and a display.
When the device is used, the foreign cotton beated out by the foreign fiber machine is manually poured into the feeding device 1 and stored in the front end cotton storage box 2, the foreign cotton is sent into the opening device 4 through the cotton conveying device 3, the feeding speed is controlled by adjusting the motor speed through the cotton conveying device 3, the opening device 4 is internally provided with a beater, an opening roller and other parts, the cotton is opened through the driving of the motor, then the opened cotton is sent into the cotton flow channel 5, the tail end of the cotton flow channel 5 is powered by the negative pressure fan component 9, so that the cotton forms cotton flow in the channel, the shooting components 6 are arranged at two sides of the cotton flow channel 5 for image detection, then the foreign fiber is beaten out through the spraying valve component 7, the impurities are taken away and enter the foreign cotton collecting component 8, finally the clean cotton is sent into the tail end cotton storage box 10, the foreign fiber detection and removal processes do not damage the cotton fiber, the cotton fiber production efficiency can be effectively improved, the labor intensity is reduced, the production cost is reduced, and the waste of raw materials is reduced.
Referring to fig. 2, a method for removing foreign fibers from raw cotton with high impurity content includes the following steps:
the method comprises the following steps: pouring the impure cotton beaten out by the foreign fiber machine into the feeding device 1 manually and storing the impure cotton in the front end cotton storage box 2;
step two: the cotton conveying device 3 conveys the miscellaneous cotton in the front end cotton storage box 2 to the opening device 4, a beater assembly and an opening roller in the opening device 4 open the miscellaneous cotton, and the opened miscellaneous cotton enters the cotton flow channel 5;
step three: the cotton flow channel 5 generates negative pressure airflow under the action of the negative pressure fan component 9 to drive the cotton impurities to slowly move towards the tail end cotton storage box 10 in the cotton flow channel 5;
step four: the miscellaneous cotton is detected through an image detection algorithm when passing through the shooting assembly 6;
step five: detecting the position of impurities in the cotton waste by an image detection algorithm, and hitting the impurities to a cotton waste collection assembly 8 when the cotton waste passes through the eruption assembly 7;
step six: the cotton flow with the impurities removed enters the tail end cotton storage box 10 for temporary storage and waits for subsequent processing.
Referring to fig. 2, a specific network structure of the image detection algorithm is shown in fig. 2, and the image detection algorithm specifically includes: the shooting component 6 shoots the cotton flow image and inputs two adjacent frame images; extracting a feature map between two frames of images through a feature extraction network and performing feature matching through a matching network; amplifying the compressed characteristic diagram to the input resolution of an original image by using an up-sampling network; the pre-labeled position information of the two is used for constraint training.
The low-order feature extractor is mainly used for extracting features such as different fiber edges, textures and color distribution characteristics, the structural design is simple, and in order to ensure the calculation efficiency of an inference end, network parameters need to meet the power series of 2. The invention designs a low-level feature extractor with three stages, the number of cores is respectively 32, 64, 128 and 256, and table 1 is a low-level feature extraction network structure.
Table 1 low order feature extraction network architecture
Figure BDA0003520335140000081
The subsequent part of the network consists of 4 residual modules, each residual module corresponds to different scales for detection, and simultaneously, all the characteristics of the previous module are fused, so that the detection capability of the small foreign fiber target is further enhanced. Each residual module consists of two groups of series convolutions, the structure of which is shown in table 2:
Figure BDA0003520335140000082
the remaining convolution part is used for up-sampling processing, the convolution kernel size is 3, the step length is 1, the padding is 1, and the band maximum pooling layer is used for down-sampling. The deconvolution parts all use the up-sampling operation of convolution kernel size 4, step length 2 and filling 1, so that the feature parameters in the feature fusion stage are kept consistent.
Compared with a classical target detection method based on a convolutional neural network, the method has the advantages that due to the fact that the resolution of an input image is fixed, too many anchor frames do not need to be designed, the anchor frames with the resolution of 1-2 times are kept on each scale, and the method can further reduce operation time.
The constraint training is modified to be rectangular frame area overlapping loss on the basis of a traditional target detection loss function, wherein the traditional target detection loss function is defined as:
Figure BDA0003520335140000091
wherein B represents the predicted number of target rectangular frames,
Figure BDA0003520335140000092
representing whether the predefined target of the image is contained currently, if it is 1, otherwise it is 0. In addition xi，yi，wi，hiIndicating the predicted coordinate information of the current target
Figure BDA0003520335140000093
The true coordinate information is predefined for the image.
The specific calculation method for the area overlapping loss of the rectangular frame is as follows:
a. defining truth coordinate information as
Figure BDA0003520335140000094
Respectively representing the initial abscissa, the ordinate, the width and the height of the rectangular frame; this prediction of coordinate information xi，yi，wi，hi；
b. For the pixel point of each position in the predicted rectangular frame, respectively calculating coordinate area information:
X＝(xi+xi+wi)*(yi+yi+hi)
Figure BDA0003520335140000101
Figure BDA0003520335140000102
Figure BDA0003520335140000103
I＝Ih*Iw
Figure BDA0003520335140000104
Figure BDA0003520335140000105
c. from the area information of the coordinates, an overlap area loss function is defined:
Figure BDA0003520335140000106
d. training a multi-scale different fiber detection network by using a new loss function;
e. and stopping training when the accuracy of the network training reaches more than 90% in the verification machine, and storing the model for deployment of the reasoning end.
In the network training, the image characteristics collected by the invention are combined, and data augmentation methods such as random brightness, random rotation, random inversion and the like are used, so that the sample diversity is increased. Because the color characteristic is mainly used for distinguishing the different fiber images from the non-different fiber images, no color augmentation method is used in the network training, and the network training is prevented from entering the unconvergeable state. In order to improve the real-time capability of the detection of the different-fiber target, the network can be changed into deep separable convolution, the parameter quantity is reduced to 1/4 of normal convolution, meanwhile, the calculation efficiency is improved by 120%, the performance is reduced by about 1.4%, and the field use requirement of the device can be still met.
In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", and the like, indicate orientations and positional relationships based on those shown in the drawings, and are used only for convenience of description and simplicity of description, and do not indicate or imply that the equipment or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be considered as limiting the present invention.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or to implicitly indicate the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (9)

1. The high-efficiency foreign fiber clearing device for the high-impurity-content raw cotton comprises an opening device (4) and is characterized in that a front end cotton storage box (2) is installed at an inlet above the opening device (4), a cotton conveying device (3) is arranged between the front end cotton storage box (2) and the opening device (4), a feeding device (1) is arranged at the top of the front end cotton storage box (2), a cotton flow channel (5) is arranged at an outlet of the opening device (4), shooting assemblies (6) are symmetrically arranged on two sides of one end, close to the opening device (4), of the cotton flow channel (5) on one side of each shooting assembly (6), impurity cotton collecting assemblies (8) and spraying valve assemblies (7) which are correspondingly arranged on two sides of the cotton flow channel (5) are installed on the cotton flow channel (5), and cotton collecting boxes (12) are installed at the tail ends of the impurity cotton collecting assemblies (8), the end of the cotton flow channel (5) is connected with a negative pressure fan assembly (9) through a pipeline, and one end of the cotton flow channel (5) close to the negative pressure fan assembly (9) is connected with an end cotton storage box (10) through a pipeline.
2. The device for removing the foreign fibers from the raw cotton with high impurity content is characterized in that a motor-driven beating assembly and a plucking roller are arranged in the plucking device (4).
3. The device for removing the foreign fibers from the raw cotton with high impurity content in the claim 1 is characterized in that a control component (11) is installed on one side of the tail end cotton storage box (10).
4. A high-efficiency foreign fiber removing device aiming at high impurity content raw cotton, according to claim 3, characterized in that the control component (11) comprises an electric control box and a display.
5. The device for removing the foreign fibers from the raw cotton with high impurity content according to claim 1, wherein the camera assembly (6) comprises a light source and an imaging sensing device.
6. The high-efficiency foreign fiber removing device aiming at the raw cotton with high impurity content is characterized in that the feeding device (1) is one of a feeding hopper, a feeding bag or a crawler feeding.
7. A high-efficiency foreign fiber removing method for raw cotton with high impurity content is characterized by comprising the following steps:
the method comprises the following steps: pouring the impure cotton beaten out by the foreign fiber machine into the feeding device (1) manually and storing the impure cotton in the front end cotton storage box (2);
step two: the cotton conveying device (3) conveys the miscellaneous cotton in the front end cotton storage box (2) to the opening device (4), a beater assembly and an opening roller in the opening device (4) open the miscellaneous cotton, and the loosened miscellaneous cotton enters the cotton flow channel (5);
step three: the cotton flow channel (5) generates negative pressure airflow under the action of the negative pressure fan component (9) to drive the miscellaneous cotton to slowly move towards the tail end cotton storage box (10) in the cotton flow channel (5);
step four: the miscellaneous cotton is detected through an image detection algorithm when passing through the shooting assembly (6);
step five: the position of impurities in the cotton waste is detected by an image detection algorithm, and the impurities are impacted to a cotton waste collecting component (8) when the cotton waste passes through the eruption component (7);
step six: the cotton flow without impurities enters a tail end cotton storage box (10) for temporary storage and waits for subsequent processing.
8. The method for removing the foreign fibers from the raw cotton with high impurity content according to claim 7, wherein the image detection algorithm specifically comprises: the shooting component (6) shoots the cotton flow image and inputs two adjacent frame images; extracting a feature map between two frames of images through a feature extraction network and performing feature matching through a matching network; amplifying the compressed characteristic diagram to the input resolution of an original image by using an up-sampling network; the pre-labeled position information of the two is used for constraint training.
9. The method for removing foreign fibers from raw cotton with high impurity content according to claim 8, wherein the constraint training is modified to rectangular frame area overlap loss based on a conventional target detection loss function, wherein the conventional target detection loss function is defined as:
Figure FDA0003520335130000031
wherein B represents the predicted number of target rectangular frames,
Figure FDA0003520335130000032
representing whether the predefined target of the image is contained currently, if it is 1, otherwise it is 0. In addition xi，yi，wi，hiIndicating the predicted coordinate information of the current target
Figure FDA0003520335130000033
The true coordinate information is predefined for the image.
The specific calculation method for the area overlapping loss of the rectangular frame is as follows:
a. defining truth coordinate information as
Figure FDA0003520335130000034
Respectively representing the initial abscissa, ordinate, rectangular frame width and rectangular frame height of the rectangular frame; this prediction of coordinate information xi，yi，wi，hi；
b. For the pixel point of each position in the predicted rectangular frame, respectively calculating coordinate area information:
X＝(xi+xi+wi)*(yi+yi+hi)
Figure FDA0003520335130000035
Figure FDA0003520335130000036
Figure FDA0003520335130000037
I＝Ih*Iw
Figure FDA0003520335130000038
Figure FDA0003520335130000039
c. from the area information of the coordinates, an overlap area loss function is defined:
Figure FDA0003520335130000041
d. training a multi-scale alien fiber detection network using a new loss function
e. And stopping training when the accuracy of the network training reaches more than 90% in the verification machine, and storing the model for deployment of the reasoning end.
CN202210176142.8A 2022-02-25 2022-02-25 Efficient foreign fiber removing device and method for high-impurity-content raw cotton Pending CN114687012A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210176142.8A CN114687012A (en) 2022-02-25 2022-02-25 Efficient foreign fiber removing device and method for high-impurity-content raw cotton

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210176142.8A CN114687012A (en) 2022-02-25 2022-02-25 Efficient foreign fiber removing device and method for high-impurity-content raw cotton

Publications (1)

Publication Number Publication Date
CN114687012A true CN114687012A (en) 2022-07-01

Family

ID=82137315

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210176142.8A Pending CN114687012A (en) 2022-02-25 2022-02-25 Efficient foreign fiber removing device and method for high-impurity-content raw cotton

Country Status (1)

Country Link
CN (1) CN114687012A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115220384A (en) * 2022-08-04 2022-10-21 武汉智目智能技术合伙企业(有限合伙) Electric control system for cotton foreign fiber machine and electric control method thereof

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN200992596Y (en) * 2006-12-30 2007-12-19 江阴市宏昌纺织机电设备有限公司 Cotton spinning unusual-fine impurity removing machine
CN201241209Y (en) * 2008-06-20 2009-05-20 南阳市利民科技开发中心 Cotton profiled fiber automatic cleaning machine
CN101979729A (en) * 2010-08-31 2011-02-23 洛阳方智测控有限公司 Seed cotton foreign fiber cleaner
CN206646207U (en) * 2017-01-17 2017-11-17 北京大恒图像视觉有限公司 A kind of storage cotton device for different fine machine and corresponding different fine machine
CN108564097A (en) * 2017-12-05 2018-09-21 华南理工大学 A kind of multiscale target detection method based on depth convolutional neural networks
CN109711288A (en) * 2018-12-13 2019-05-03 西安电子科技大学 Remote sensing ship detecting method based on feature pyramid and distance restraint FCN
CN111582257A (en) * 2019-02-15 2020-08-25 波音公司 Method, device and system for detecting object to be detected
CN111636120A (en) * 2020-01-20 2020-09-08 武汉裕大华纺织有限公司 High-performance foreign fiber sorting device and sorting method
CN211921768U (en) * 2019-12-05 2020-11-13 北京经纬纺机新技术有限公司 Foreign fiber detection system based on high-performance display card
CN112101134A (en) * 2020-08-24 2020-12-18 深圳市商汤科技有限公司 Object detection method and device, electronic device and storage medium
CN113409250A (en) * 2021-05-26 2021-09-17 杭州电子科技大学 Solder joint detection method based on convolutional neural network

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN200992596Y (en) * 2006-12-30 2007-12-19 江阴市宏昌纺织机电设备有限公司 Cotton spinning unusual-fine impurity removing machine
CN201241209Y (en) * 2008-06-20 2009-05-20 南阳市利民科技开发中心 Cotton profiled fiber automatic cleaning machine
CN101979729A (en) * 2010-08-31 2011-02-23 洛阳方智测控有限公司 Seed cotton foreign fiber cleaner
CN206646207U (en) * 2017-01-17 2017-11-17 北京大恒图像视觉有限公司 A kind of storage cotton device for different fine machine and corresponding different fine machine
CN108564097A (en) * 2017-12-05 2018-09-21 华南理工大学 A kind of multiscale target detection method based on depth convolutional neural networks
CN109711288A (en) * 2018-12-13 2019-05-03 西安电子科技大学 Remote sensing ship detecting method based on feature pyramid and distance restraint FCN
CN111582257A (en) * 2019-02-15 2020-08-25 波音公司 Method, device and system for detecting object to be detected
CN211921768U (en) * 2019-12-05 2020-11-13 北京经纬纺机新技术有限公司 Foreign fiber detection system based on high-performance display card
CN111636120A (en) * 2020-01-20 2020-09-08 武汉裕大华纺织有限公司 High-performance foreign fiber sorting device and sorting method
CN112101134A (en) * 2020-08-24 2020-12-18 深圳市商汤科技有限公司 Object detection method and device, electronic device and storage medium
CN113409250A (en) * 2021-05-26 2021-09-17 杭州电子科技大学 Solder joint detection method based on convolutional neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
JIAHUI YU 等: "UnitBox: An Advanced Object Detection Network", 《HTTPS://ARXIV.ORG/PDF/1608.01471.PDF》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115220384A (en) * 2022-08-04 2022-10-21 武汉智目智能技术合伙企业(有限合伙) Electric control system for cotton foreign fiber machine and electric control method thereof

Similar Documents

Publication Publication Date Title
CN104737707B (en) A kind of combined harvester cleans percentage of impurity adaptive controller and adaptive cleaning method
CN104138851B (en) Vibrating feeding and imaging recognition-based large particle material sorting system and method
CN101570905B (en) Seed cotton foreign fiber separator
CN103211292B (en) Equipment and method for sectional fine sorting of tobacco sheets
CN114687012A (en) Efficient foreign fiber removing device and method for high-impurity-content raw cotton
CN204741825U (en) Combine harvester cleans dirt percentage adaptive control device
CN1485616A (en) Fowl eggs quality non-destruction automatic detection grading apparatus and process
CN111636122B (en) Bale plucking impurity removing device and removing method
CN106370667A (en) Visual detection apparatus and method for quality of corn kernel
CN102660815A (en) Cotton foreign fiber automatic removing method and system based on image acquisition and DSP (digital signal processing) algorithm
CN109030504A (en) The panorama potato defect detecting device and method combined based on actual situation imaging
CN110359132B (en) Novel cotton grabbing process suitable for intelligent spinning
CN208098653U (en) A kind of unginned cotton residual film sorting unit
CN112730417B (en) Online detection device for appearance quality of wheat
CN207681044U (en) A kind of ganoderma lucidum automatic sorting apparatus
CN211652867U (en) Device for rapidly evaluating impurities of seed cotton in purchasing link
CN106723322B (en) A kind of thin cigarette pipe tobacco of cigarette machine defective cigarette returns online mixes device
CN203565315U (en) Drum sieve used for compound cleaning of grains
CN111353432A (en) Rapid honeysuckle medicinal material cleaning method and system based on convolutional neural network
KR101800252B1 (en) Pre-processing apparatus of sea-squirt harvest for an automatic sorting of sea-squirt
CN214794500U (en) Online detection device for wheat appearance quality
CN205650590U (en) Old and useless panel face clearance production line
CN111624948B (en) Intelligent cotton grabbing management system
CN221311556U (en) Sorting unit of all-round outer quality detection of fruit
CN211538634U (en) Full-automatic removing machine

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20220701