Cotton detection device and method based on deep learning

Abstract

The invention provides a cotton detection device and method based on deep learning, and relates to the technical field of cotton detection. In order to solve the problems of low portability and low detection efficiency of a detection device in the prior art, an intelligent light supplementing module is integrated on an image acquisition module, and the intelligent light supplementing module is used for supplementing light according to the light intensity of a target detection environment; the image acquisition module comprises a mobile terminal and a fixing structure, and the fixing structure is used for fixing the mobile terminal to acquire images according to a set shooting view angle; the image detection module is in communication connection with the image acquisition module, a deep learning grading model is embedded in the image detection module, and the deep learning grading model is used for extracting key features of an input image, analyzing and processing the key features and then outputting a grading result; the image detection module further comprises a mobile interaction interface and is further used for transmitting the grading result to the mobile terminal through the mobile interaction interface. The detection device disclosed by the invention is relatively high in portability and detection efficiency.

Description

Technical Field

[0001] This invention relates to the field of cotton detection technology, and in particular to a cotton detection device and method based on deep learning. Background Technology

[0002] Cotton testing is a crucial step in determining the grade of cotton based on core quality indicators using professional technical means. The grading and testing of imported cotton directly impacts textile companies' control over raw material quality; accurate and efficient testing can shorten customs clearance cycles and reduce logistics costs for businesses.

[0003] Current cotton testing technology has evolved from traditional manual grading to automated testing, generally including large-scale laboratory testing devices and small to medium-sized automated testing devices. Small to medium-sized automated testing devices are mostly based on machine vision and image processing technologies, relying on dedicated image acquisition devices and supporting lighting components, such as industrial-grade cameras and fixed-focus optical lenses. While this type of hardware can meet the required accuracy, its large size and weight result in poor portability and low testing efficiency.

[0004] Therefore, developing a cotton detection device and method based on deep learning is of great significance for improving the portability and detection efficiency of the detection device. Summary of the Invention

[0005] To address the issues of portability and low detection efficiency in existing detection devices, this invention proposes a cotton detection device based on deep learning, specifically comprising:

[0006] Intelligent supplementary lighting module, image acquisition module, and image detection module;

[0007] The intelligent supplementary lighting module is integrated on the image acquisition module, and the intelligent supplementary lighting module is used to supplement the light according to the light intensity of the target detection environment;

[0008] The image acquisition module includes a mobile terminal and a fixed structure, wherein the fixed structure is used to fix the mobile terminal to acquire images according to a set shooting angle;

[0009] The image detection module is communicatively connected to the image acquisition module. The image detection module embeds a deep learning hierarchical model, which is used to extract key features of the input image and output hierarchical results after analyzing and processing the key features.

[0010] The image detection module also includes a mobile interaction interface, and the image detection module is further used to transmit the grading results to the mobile terminal through the mobile interaction interface.

[0011] Furthermore, the intelligent supplementary lighting module includes a light source component and a control unit. The light source component is connected to the control unit, which is used to detect the light intensity of the target detection environment and control the light source component to supplement light when the light intensity is less than a set light intensity.

[0012] Furthermore, the light source component includes an array of light-emitting diodes (LEDs) that are uniformly distributed around the camera device of the mobile terminal.

[0013] Furthermore, the image detection module also includes a processing unit, which is used to perform standardization preprocessing on the input image to obtain a standard image.

[0014] Furthermore, the deep learning grading model is connected to the processing unit. The deep learning grading model is used to extract key features of the standard image, and outputs grading results after analyzing and processing the key features.

[0015] Furthermore, the key features include color features, impurity features, and rolling quality features.

[0016] Furthermore, the fixing structure includes an adjustable clamping component and a height positioning bracket; the adjustable clamping component is fixedly connected to the height positioning bracket, the adjustable clamping component is used to adapt to mobile terminals of different sizes, and the height positioning bracket can be adjusted vertically to ensure that the acquisition distance meets the set requirements.

[0017] The present invention also provides a cotton detection method based on deep learning, applicable to the cotton detection device based on deep learning described in any of the above claims, the method comprising the following steps:

[0018] The intelligent fill light module is integrated into the image acquisition module;

[0019] The intelligent supplemental lighting module is controlled to provide supplemental lighting based on the light intensity of the target detection environment;

[0020] Control the mobile terminal in the image acquisition module to acquire images;

[0021] The deep learning hierarchical model in the image detection module extracts key features from the input image, analyzes and processes the key features, and outputs hierarchical results.

[0022] The grading results are transmitted to the mobile terminal via the mobile interactive interface.

[0023] Furthermore, the training process of the deep learning hierarchical model includes: taking the key features as input and the standard level as output to train the deep learning hierarchical model.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0025] This invention provides a deep learning-based cotton inspection device, comprising: an intelligent supplementary lighting module, an image acquisition module, and an image detection module. The intelligent supplementary lighting module is integrated into the image acquisition module and is used to supplement lighting according to the light intensity of the target inspection environment. The image acquisition module includes a mobile terminal and a fixed structure, the fixed structure being used to fix the mobile terminal to acquire images according to a set shooting angle. The image detection module is communicatively connected to the image acquisition module, and embeds a deep learning grading model within the image detection module. The deep learning grading model is used to extract key features from the input image, analyze and process the key features, and output grading results. The image detection module also includes a mobile interaction interface, which is used to transmit the grading results to the mobile terminal. By integrating the intelligent supplementary lighting module into the image acquisition module, and combining it with a mobile terminal and a fixed structure to achieve standardized image acquisition, the device eliminates the dependence on laboratory-specific lighting and fixed locations required for traditional cotton grading inspection. Inspectors can conduct grading inspections of imported cotton anytime, anywhere, making the inspection device more portable and significantly shortening the inspection cycle while improving inspection efficiency.

[0026] Furthermore, by embedding a deep learning grading model into the image detection module, the detection device extracts key image features and outputs grading results. This eliminates interference from subjective judgment and visual fatigue inherent in manual grading, improving the accuracy and impartiality of the detection results and reducing the re-inspection rate. Simultaneously, the grading results can be transmitted to mobile terminals, adapting to the needs of port scenarios, simplifying the detection process, lowering the professional threshold, and effectively improving the efficiency of customs cotton inspection and supervision. Attached Figure Description

[0027] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the structure of a cotton detection device based on deep learning provided in an embodiment of the present invention;

[0029] Figure 2 This is a flowchart of a cotton detection method based on deep learning provided by an embodiment of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0031] The specific embodiments of the present invention will be described below.

[0032] To address the issues of low portability and detection efficiency in existing detection devices, this invention integrates an intelligent supplementary lighting module into the image acquisition module. This module provides supplementary lighting based on the ambient light intensity of the target detection environment. The image acquisition module includes a mobile terminal and a fixed structure. The fixed structure secures the mobile terminal to capture images from a set shooting angle. An image detection module is communicatively connected to the image acquisition module. This module embeds a deep learning hierarchical model to extract key features from the input image, analyzes these features, and outputs a hierarchical result. The image detection module also includes a mobile interaction interface, which transmits the hierarchical result to the mobile terminal. This invention provides a detection device with high portability and detection efficiency.

[0033] Example 1

[0034] This invention provides a method for urban rail transit engineering model design based on collaborative management of design data. Figure 1 This is a schematic diagram of a cotton detection device based on deep learning provided in an embodiment of the present invention, as shown below. Figure 1 As shown, the cotton detection device includes an intelligent supplementary lighting module 110, an image acquisition module 120, and an image detection module 130. The intelligent supplementary lighting module 110 is integrated into the image acquisition module 120 and is used to supplement light according to the light intensity of the target detection environment. The image acquisition module 120 includes a mobile terminal 121 and a fixed structure 122, which is used to fix the mobile terminal 121 to acquire images according to a set shooting angle. The image detection module 130 is communicatively connected to the image acquisition module 120. The image detection module 130 embeds a deep learning grading model 131, which is used to extract key features from the input image and output grading results after analyzing and processing the key features. The image detection module 130 also includes a mobile interaction interface 132, which is used to transmit the grading results to the mobile terminal 121.

[0035] The intelligent supplementary lighting module 110 is an adaptive lighting module for different detection environments, capable of adjusting color, illuminance, and illuminance uniformity, and meeting the standard technical requirements of cotton-specific grading rooms. The image acquisition module 120 acquires cotton feature images, and the mobile terminal 121 is the core acquisition terminal for these images, also used to receive grading results. For example, the mobile terminal can be a compact and portable device such as a mobile phone or tablet, meeting pixel and accuracy requirements. The fixing structure 122 is used to fix the mobile terminal, allowing for adjustment of the shooting angle and distance, ensuring standardized image acquisition. The image detection module 130 analyzes and processes the input image and outputs grading results. The deep learning grading model 131 is an intelligent algorithm model trained based on the input image and standard grades; the input image can be a sawtooth cotton image. The mobile interaction interface 132 is the interface between the image detection module and the mobile terminal for transmitting grading results, ensuring the results can be displayed on the mobile device.

[0036] The intelligent supplementary lighting module is integrated into the image acquisition module, achieving a unified lighting system and acquisition terminal for convenient portability. The intelligent supplementary lighting module adjusts the light source according to the ambient light intensity, ensuring the lighting in the acquisition environment matches the standard requirements for cotton grading detection. A fixed structure limits the position and viewing angle of the mobile terminal, ensuring the standardization and uniformity of image acquisition. The mobile terminal captures cotton samples from the set viewing angle, obtaining feature images that meet the model's recognition requirements. A data transmission link is established between the image acquisition module and the image detection module, enabling the exchange of acquired images. The trained deep learning grading model is implanted into the image detection module as the core of image analysis and processing. The deep learning grading model identifies and extracts key features from the input image, performs calculations and judgments on the extracted key features to match the cotton grading standards. The image detection module generates the cotton grading results and sends them to the mobile terminal via a mobile interactive interface for easy viewing. The mobile terminal receiving the grading results can be distinct from the mobile terminal in the image acquisition module.

[0037] This embodiment provides a deep learning-based cotton inspection device, comprising: an intelligent supplementary lighting module, an image acquisition module, and an image detection module. The intelligent supplementary lighting module is integrated into the image acquisition module and is used to supplement lighting according to the light intensity of the target inspection environment. The image acquisition module includes a mobile terminal and a fixed structure, with the fixed structure used to fix the mobile terminal to acquire images according to a set shooting angle. The image detection module is communicatively connected to the image acquisition module and embeds a deep learning grading model. The deep learning grading model is used to extract key features from the input image and output grading results after analyzing and processing the key features. The image detection module also includes a mobile interaction interface, which is used to transmit the grading results to the mobile terminal. By integrating the intelligent supplementary lighting module into the image acquisition module, and using a mobile terminal and a fixed structure to achieve standardized image acquisition, the device eliminates the dependence on laboratory-specific lighting and fixed locations required for traditional cotton grading inspection. Inspectors can conduct grading inspections of imported cotton anytime and anywhere, making the inspection device more portable and significantly shortening the inspection cycle while improving inspection efficiency.

[0038] Furthermore, by embedding a deep learning grading model into the image detection module, the detection device extracts key image features and outputs grading results. This eliminates interference from subjective judgment and visual fatigue inherent in manual grading, improving the accuracy and impartiality of the detection results and reducing the re-inspection rate. Simultaneously, the grading results can be transmitted to mobile terminals, adapting to the needs of port scenarios, simplifying the detection process, lowering the professional threshold, and effectively improving the efficiency of customs cotton inspection and supervision.

[0039] Based on the above embodiments, the intelligent supplementary lighting module 110 includes a light source component and a control unit. The light source component is connected to the control unit, which is used to detect the light intensity of the target detection environment and control the light source component to supplement light when the light intensity is less than the set light intensity.

[0040] The light source component is the core component of the intelligent supplemental lighting module, providing supplemental lighting and meeting colorimetric and illuminance requirements. The control unit is the core control component of the intelligent supplemental lighting module, responsible for detecting ambient light intensity, adjusting the brightness of the light source component based on the light intensity value, and controlling the start and stop of the light source component.

[0041] The control unit detects the light intensity of the target detection environment and compares the detected value with the set light intensity that meets the cotton grading standard. If the detected light intensity is lower than the set light intensity, the control unit sends a command to the connected light source component to control the light source component to start and supplement the illumination so that the ambient light intensity meets the cotton grading detection requirements.

[0042] The intelligent supplemental lighting module provided in this embodiment automates light detection and supplemental lighting without manual adjustment, reducing reliance on the operator's experience. The ambient light intensity after supplemental lighting matches the standards of a dedicated grading room, eliminating interference from lighting differences in different scenarios and ensuring the standardization of cotton image acquisition. The intelligent supplemental lighting module can adapt to lighting conditions in non-laboratory scenarios such as border crossings, overcoming the limitations of dedicated laboratory lighting and enabling flexible testing in multiple scenarios.

[0043] Specifically, the light source component includes a light-emitting diode (LED) array, which is uniformly distributed around the camera on the mobile terminal. The LED array, a collection of multiple LEDs, provides a more stable and wider-coverage supplementary lighting effect. The camera on the mobile terminal is the core hardware for capturing images of the cotton sample.

[0044] By evenly distributing the LED array around the imaging device, the lighting coverage area can be made more uniform, avoiding areas that are too bright, too dark, or have blind spots. This ensures consistent lighting conditions for the captured cotton images, eliminates image deviations caused by uneven lighting, and improves the accuracy of subsequent deep learning grading models.

[0045] Based on the above embodiments, the image detection module further includes a processing unit, which is used to perform standardization preprocessing on the input image to obtain a standard image.

[0046] The processing unit is a unit that performs standardization preprocessing on the input cotton image. Standardization preprocessing is a unified processing operation on the original cotton image. Through standardization preprocessing, background interference of the input image can be eliminated, and the image size can be unified so that it matches the recognition requirements of the deep learning hierarchical model.

[0047] A deep learning grading model is connected to the processing unit. This model extracts key features from the standard image, analyzes and processes these features, and outputs a grading result. The key features include color features, impurity features, and rolling quality features.

[0048] After the processing unit performs standardized preprocessing on the input image and generates a standard image, it outputs the standard image to the deep learning grading model. The deep learning grading model first extracts three key features from the standard image: color, impurities, and ginning quality. Then, it analyzes and processes these features to match the cotton grading standards and finally outputs the corresponding cotton grading results.

[0049] The deep learning-based grading model focuses on the core judgment dimensions of cotton grade, including color features, impurity features, and ginning quality features. This targeted approach improves the accuracy of grading results and reduces judgment errors caused by missing features. Furthermore, based on features extracted from standardized images, it eliminates problems such as background interference and uneven lighting in the original images, resulting in a high degree of consistency between the model's recognition results and manual grading, thus reducing the need for re-inspection.

[0050] Based on the above embodiments, the fixed structure includes an adjustable clamping component and a height positioning bracket; the adjustable clamping component is fixedly connected to the height positioning bracket, the adjustable clamping component is used to adapt to mobile terminals of different sizes, and the height positioning bracket can be adjusted vertically to ensure that the acquisition distance meets the set requirements.

[0051] The adjustable clamping assembly is a clamping component that can be adjusted to match the size of mobile terminals of different dimensions, exhibiting strong adaptability. The height positioning bracket is a bracket that adjusts the height vertically; its core function is to adjust the acquisition distance between the mobile terminal and the cotton sample. The acquisition distance, the distance between the mobile terminal's imaging device and the cotton sample, is a key parameter ensuring that the acquired images meet laboratory standards.

[0052] The fixed structure consists of an adjustable clamping component and a height positioning bracket. In use, the adjustable clamping component is first used to adapt and fix mobile terminals of different sizes. Then, according to the set distance requirements for cotton image acquisition, the height of the height positioning bracket is adjusted vertically to ensure that the acquisition distance between the mobile terminal and the cotton sample meets the standardized acquisition requirements.

[0053] By incorporating adjustable clamping components to accommodate mobile terminals of different sizes, the versatility of the fixed structure is enhanced. This eliminates the need for separately designed fixing components for different specifications of acquisition terminals, reducing usage costs. The fixed structure, combined with a portable mobile terminal, makes the image acquisition module compact and flexible in use, freeing it from the limitations of fixed acquisition conditions in laboratories. It enables standardized acquisition of cotton images in various scenarios, such as border crossings.

[0054] The deep learning-based cotton inspection device provided in this embodiment integrates an intelligent supplementary lighting module into the image acquisition module. Combined with a mobile terminal and a fixed structure, it achieves standardized image acquisition, eliminating the reliance on laboratory lighting and fixed locations required for traditional cotton grading inspections. Inspectors can conduct grading inspections of imported cotton anytime, anywhere, making the device more portable and significantly shortening the inspection cycle while improving efficiency. The intelligent supplementary lighting module automates illumination detection and supplementation, eliminating the need for manual adjustment and reducing reliance on the operator's experience. The ambient light intensity after supplementation matches the standards of a dedicated grading room, eliminating interference from lighting differences in different scenarios and ensuring the standardization of cotton image acquisition. The intelligent supplementary lighting module can adapt to the lighting conditions of non-laboratory scenarios such as border crossings, overcoming the limitations of laboratory lighting and enabling flexible inspection in multiple scenarios. The deep learning grading model focuses on the core judgment dimensions of cotton grade: color characteristics, impurity characteristics, and ginning quality characteristics, providing strong targeting and improving the accuracy of grading results, reducing judgment errors caused by missed features.

[0055] Example 2

[0056] This invention also provides a cotton detection method based on deep learning. Figure 2 This is a flowchart of a cotton detection method based on deep learning provided by an embodiment of the present invention, as shown below. Figure 2 As shown, the method includes the following steps:

[0057] S1. The intelligent supplementary lighting module is integrated into the image acquisition module.

[0058] S2. Control the intelligent supplementary lighting module to supplement light according to the light intensity of the target detection environment.

[0059] S3. Control the mobile terminal in the image acquisition module to acquire images.

[0060] S4. The deep learning hierarchical model in the image detection module extracts key features of the input image, analyzes and processes the key features, and outputs the hierarchical results.

[0061] S5. The grading results are transmitted to the mobile terminal through the mobile interactive interface.

[0062] The training process of the deep learning hierarchical model includes: taking the key features as input and the standard level as output to train the deep learning hierarchical model.

[0063] An intelligent supplementary lighting module is integrated into the image acquisition module. This module adjusts the light source according to the ambient light intensity, ensuring the ambient illumination matches the standard requirements for cotton grading. A fixed structure defines the position and viewing angle of the mobile terminal, controlling it to capture images of cotton samples from the set angle, obtaining feature images that meet the model's recognition requirements. A trained deep learning grading model is then embedded into the image detection module, serving as the core of image analysis and processing. This model identifies and extracts key features from the input image, performing calculations and judgments on these features to match the cotton grading standards. The image detection module generates the cotton grading results and sends them to the mobile terminal via a mobile interface.

[0064] The deep learning-based cotton detection method provided in this embodiment is executed by the deep learning-based cotton detection device provided in any of the above embodiments, and has the beneficial effects of any of the above embodiments, which will not be repeated here.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A cotton detection device based on deep learning, characterized in that, The application relates to a deep learning-based cotton detection device. The intelligent light supplementing module is integrated on the image acquisition module, and is used for light supplementing according to the light intensity of a target detection environment. The image acquisition module comprises a mobile terminal and a fixed structure, and the fixed structure is used for fixing the mobile terminal to collect images according to a set shooting angle. The image detection module is in communication connection with the image acquisition module, and a deep learning hierarchical model is embedded in the image detection module. The image detection module further comprises a mobile interactive interface, and the image detection module is further used for transmitting the hierarchical result to the mobile terminal through the mobile interactive interface. The intelligent light supplementing module comprises a light source assembly and a control unit, the light source assembly is connected with the control unit, the control unit is used for detecting the light intensity of a target detection environment, and when the light intensity is less than a set light intensity, the control unit controls the light source assembly to supplement light.

2. The deep learning-based cotton detection device of claim 1, wherein, The light source assembly comprises a light emitting diode array, and the light emitting diode array is uniformly distributed around the shooting device of the mobile terminal.

3. The deep learning-based cotton detection device of claim 2, wherein, The image detection module further comprises a processing unit, and the processing unit is used for standardizing and pretreating input images to obtain standard images.

4. The deep learning-based cotton detection device of claim 1, wherein, 5. The deep learning-based cotton detection device according to claim 4, wherein The deep learning hierarchical model is connected with the processing unit, and the deep learning hierarchical model is used for extracting key features of the standard images and outputting hierarchical results after analyzing and processing the key features. The key features comprise color features, impurity features and rolling quality features.

6. The deep learning-based cotton detection device of claim 5, wherein, The fixed structure comprises an adjustable clamping assembly and a height positioning support, the adjustable clamping assembly is fixedly connected with the height positioning support, the adjustable clamping assembly is used for adapting to mobile terminals of different sizes, and the height positioning support can be adjusted in the vertical direction to adjust the height so that the collection distance meets the set requirement.

7. The deep learning-based cotton detection device of claim 1, wherein, The detection method is executed by the deep learning-based cotton detection device according to any one of claims 1-7, and the method comprises the following steps:

8. A deep learning-based cotton detection method, characterized in that, The intelligent light supplementing module is integrated on the image acquisition module; The intelligent light supplementing module is controlled to supplement light according to the light intensity of a target detection environment; The mobile terminal in the image acquisition module is controlled to collect images; The deep learning hierarchical model in the image detection module extracts key features of input images and outputs hierarchical results after analyzing and processing the key features; The hierarchical result is transmitted to the mobile terminal through the mobile interactive interface. The training process of the deep learning hierarchical model comprises:

9. The deep learning-based cotton detection method of claim 8, wherein, The key features are taken as input, and standard grades are taken as output, so that the deep learning hierarchical model is trained. ​

Images