Joint excitation up-sampling network calculation method and detection system for cigarette ash packing crack segmentation

Abstract

The invention provides a combined excitation up-sampling network calculation method and detection system for cigarette ash packing crack segmentation, and belongs to the technical field of cigarette detection. The combined excitation up-sampling network calculation method comprises the following steps: acquiring a crack image in a cigarette combustion process, and preprocessing the crack image; performing multi-stage feature extraction on the preprocessed crack image by adopting an encoder, obtaining feature maps of different scales through convolution and pooling operation of four stages, and performing down-sampling through 2 * 2 maximum pooling between every two stages; a combined excitation up-sampling module is introduced into the decoder, and the combined excitation up-sampling module comprises a space attention mechanism and a combined excitation channel mechanism; the joint excitation up-sampling module is used for carrying out joint up-sampling on the feature maps with different scales to the same size by utilizing multi-stage feature map joint up-sampling, and carrying out cascade processing to generate a feature map containing multi-scale information.

Description

Technical Field

[0001] The present invention relates to the technical field of cigarette detection, and in particular to a joint excitation up-sampling network calculation method and a detection system for cigarette ash crack segmentation. Background Art

[0002] The two important parameters of cigarette ash quality are the shrinkage of the ash column and the proportion of ash cracks. Among them, the shrinkage of the ash column mainly reflects the filling rate of tobacco. The internal pores of cigarettes with high filling rates are small, and the ash column shrinks less after burning. The formation of ash cracks is caused by multiple factors such as the wrapping performance of cigarette paper, tobacco formula, filling performance, and combustion thermal performance. The proportion of cracks on the ash column after burning is positively correlated with the ash flakes scattered during smoking. The ash flakes scattered will cause environmental pollution on the one hand, and affect consumers' smoking experience on the other hand.

[0003] Traditional cigarette pack ash quality detection methods usually rely on manual detection or simple visual image processing technology. For visual image detection methods, the detection of ash column shrinkage is relatively easy to achieve, however, the accurate detection of crack area ratio has always been a challenging problem. On the one hand, there are many types of cigarette paper colors and textures, and the surface morphology after burning is relatively complex. On the other hand, the inadequate burning of cigarettes will lead to large differences in the color of the ash column of the pack (yellow, black, gray, white), and the contrast between the cracks at different positions and the background is different. Conventional image segmentation algorithms are difficult to accurately extract the crack area.

[0004] In view of this, the inventor, based on many years of production design experience in this field and related fields, has designed a joint excitation upsampling network calculation method and detection system for cigarette ash crack segmentation after repeated experiments, in order to solve the problems existing in the prior art. Summary of the invention

[0005] The object of the present invention is to provide a joint excitation upsampling network calculation method and detection system for cigarette ash crack segmentation, which can effectively improve the segmentation accuracy of cigarette ash cracks.

[0006] To achieve the above-mentioned object of the invention, the present invention proposes a joint excitation upsampling network calculation method and a detection system for cigarette ash crack segmentation, wherein the joint excitation upsampling network calculation method comprises:

[0007] Acquire crack images during cigarette burning and pre-process the crack images;

[0008] The encoder is used to perform multi-level feature extraction on the preprocessed crack image. Feature maps of different scales are obtained through four stages of convolution and pooling operations. Downsampling is performed between each two stages through 2×2 maximum pooling.

[0009] A joint excitation upsampling module is introduced in the decoder, which includes a spatial attention mechanism and a joint excitation channel mechanism.

[0010] The joint excitation upsampling module uses multi-level feature maps to jointly upsample, upsamples feature maps of different scales to the same size and performs cascade processing to generate feature maps containing multi-scale information.

[0011] The present invention also provides a detection system for cigarette ash cracks, characterized in that the detection system implements the above method, including:

[0012] An image acquisition module, used for real-time acquisition of crack image data during cigarette burning;

[0013] A data preprocessing module is used to remove noise, calibrate grayscale and enhance contrast of image data;

[0014] The segmentation module includes the JEUNet encoder-decoder structure and the joint excitation upsampling module, which combines the upsampling operation of the multi-level feature map to achieve accurate segmentation of the crack area;

[0015] The output module is used to generate visual segmentation results and analysis reports of the crack area.

[0016] The present invention also provides a computer-readable storage medium on which a computer program is stored, characterized in that the computer program implements the steps of the above-mentioned calculation method when executed by a processor.

[0017] The present invention also proposes a computer device, comprising a memory and a processor, wherein a computer program that can be run on the processor is stored in the memory, and the device is characterized in that the steps of the above-mentioned calculation method are implemented when the processor executes the computer program.

[0018] Compared with the prior art, the present invention has the following characteristics and advantages:

[0019] The present invention proposes a joint excitation up-sampling network calculation method and detection system for cigarette ash crack segmentation. By designing an encoder-decoder symmetric structure and introducing a joint excitation up-sampling module in the decoder part, the joint up-sampling of multi-scale feature maps and space-channel excitation are realized, thereby improving the segmentation accuracy and efficiency of cigarette cracks. The joint excitation up-sampling network calculation method proposed in the present invention enhances the segmentation accuracy of the crack area through the spatial attention mechanism and the channel excitation mechanism, while reducing the computational complexity and improving the algorithm efficiency. In addition, the Binary Cross-Entropy loss function is used for model optimization, and the 10-fold Cross Validation method is used to improve the robustness and generalization ability of the model. While ensuring the high-precision crack segmentation effect, the computing resource requirements are significantly reduced, and the multi-scale perception ability and segmentation accuracy of the segmentation model are improved, providing an efficient and accurate solution for cigarette crack detection, which has broad industrial application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings described herein are only for explanation purposes and are not intended to limit the scope of the present invention in any way. In addition, the shapes and proportional dimensions of the various components in the figures are only schematic, used to help understand the present invention, and are not specifically limited to the shapes and proportional dimensions of the various components of the present invention. Those skilled in the art can select various possible shapes and proportional dimensions to implement the present invention according to the teachings of the present invention.

[0021] Figure 1 It is a schematic diagram of the JEUNet network structure of the present invention;

[0022] Figure 2 This is a schematic diagram of the structure of the joint excitation upsampling (JEU) module of the present invention;

[0023] Figure 3 A schematic diagram showing the comparison of the accuracy and computational complexity of JEUNet of the present invention and other advanced segmentation algorithms;

[0024] Figure 4 Schematic diagram of the comparison of cigarette ash crack map, ground truth and multi-method segmentation effects of the present invention

[0025] Figure 5 It is a schematic diagram of the steps of the joint excitation upsampling network calculation method for cigarette ash crack segmentation of the present invention. DETAILED DESCRIPTION

[0026] The details of the present invention can be more clearly understood with reference to the accompanying drawings and the description of the specific embodiments of the present invention. However, the specific embodiments of the present invention described herein are only used for the purpose of explaining the present invention and cannot be understood as limiting the present invention in any way. Under the guidance of the present invention, technicians can conceive of any possible variations based on the present invention, which should all be considered to belong to the scope of the present invention.

[0027] The present invention proposes a joint excitation upsampling network calculation method for cigarette ash crack segmentation, such as Figure 5 As shown, the joint excitation upsampling network calculation method includes:

[0028] Acquire crack images during cigarette burning and pre-process the crack images;

[0029] The encoder is used to perform multi-level feature extraction on the preprocessed crack image. Feature maps of different scales are obtained through four stages of convolution and pooling operations. Downsampling is performed between each two stages through 2×2 maximum pooling.

[0030] A joint excitation upsampling module is introduced in the decoder, which includes a spatial attention mechanism and a joint excitation channel mechanism.

[0031] The joint excitation upsampling module uses multi-level feature maps to jointly upsample, upsamples feature maps of different scales to the same size and performs cascade processing to generate feature maps containing multi-scale information.

[0032] The present invention proposes a joint excitation upsampling network calculation method for cigarette ash crack segmentation. First, the preprocessing step ensures the image quality and provides high-quality input for feature extraction. Secondly, the multi-level feature extraction capability of the encoder enables the model to capture multi-scale information from shallow to deep layers, which is crucial for the accurate identification of crack areas. The spatial attention mechanism and joint excitation channel mechanism of the joint excitation upsampling module further enhance the model's recognition ability of the crack area, and achieves a refined response to the crack area features by adaptively recalibrating the feature response in the channel direction and modeling the interdependence between different channels. Finally, the joint upsampling and cascade processing of the multi-level feature maps generates a feature map containing multi-scale information, which not only improves the multi-scale perception capability of the segmentation model, but also improves the accuracy and upsampling speed of crack segmentation.

[0033] In an optional embodiment of the present invention, the spatial attention mechanism processes the feature map through maximum pooling and average pooling operations. The technical effect of the present invention is that the accuracy and efficiency of cigarette ash crack segmentation are significantly improved. Through the spatial attention mechanism, the model can more accurately identify the crack area, especially the significance under complex backgrounds. The maximum pooling and average pooling operations enable the feature map to enhance the recognition ability of the crack boundary while retaining important spatial information, thereby improving the accuracy of segmentation.

[0034] In an optional embodiment of the present invention, the joint excitation channel mechanism generates a channel excitation vector through global average pooling, and adjusts the weights of the feature responses of different channels. By introducing the joint excitation channel mechanism, the calculation method of the present invention achieves a technical breakthrough in the task of cigarette ash crack segmentation. The channel excitation vector generated by global average pooling enables the model to capture the importance of each channel and strengthens the feature response through weight adjustment.

[0035] In an optional example of this implementation, multi-level feature map joint upsampling is used, including: sampling is doubled upsampling through 2×2 deconvolution operation, and each upsampled feature map is cascaded to generate a multi-level feature map containing different receptive fields to improve the multi-scale perception ability of the segmentation model; after the multi-level feature map is jointly upsampled, the feature map size is adjusted to the original image size through 1×1 convolution. By adopting the technical solution of multi-level feature map joint upsampling, while improving the segmentation accuracy, the computational efficiency is also maintained, providing strong technical support for the efficient and accurate detection of cigarette ash cracks.

[0036] Specifically, packet crack detection can be described as a pixel-level anomaly prediction task. The proposed joint excitation upsampling network JEUNet has a classic encoder-decoder symmetric structure. Its network framework is as follows: Figure 1 As shown. The yellow arrow indicates direct input; the black arrow indicates a 3×3 convolution operation with a stride of 1, which uses the zero-padding strategy to keep the feature maps of the same level at the same size; the gray arrow indicates a cut and connect operation, which cuts the left feature map to the size corresponding to the right and then connects it; the blue arrow indicates a 1×1 operation for final classification, and the last two layers of output are the result and the background; the red arrow indicates downsampling the feature map by 2×2 maximum pooling; the green arrow indicates upsampling the feature map by a factor of 2 using a 2×2 deconvolution operation. k indicates the number of base channels of the convolution feature map (in this article, k = 32).

[0037] In an optional embodiment of the present invention, a channel excitation path is embedded in the multi-level feature map of the second stage, the third stage and the fourth stage of the decoding stage, and the multi-scale crack features are gradually refined through the channel excitation path, and the dependency relationship between multiple channel features is modeled. By embedding the channel excitation path in the decoding stage, not only the recognition ability of the model for multi-scale crack features is enhanced, but also the representation of the feature map is optimized by modeling the dependency relationship between channels, making the segmentation result more accurate. The introduction of the channel excitation path enables the model to adaptively adjust the feature responses of different channels, so as to better capture the detailed information of the crack area. Its step-by-step refinement of feature processing and modeling of the dependency relationship between channels provide strong technical support for the accurate segmentation of cigarette ash cracks, and improve the robustness and generalization ability of the segmentation model.

[0038] Specifically, the encoder part of JEUNet is consistent with the backbone network of UNet. The original input image is encoded in 4 stages through continuous convolutional layers, and is transmitted through 2× maximum pooling downsampling between every 2 stages. Therefore, there are 5 scales including the original image. The feature map of each scale contains information with different receptive fields. The shallow feature map is mainly the detailed texture information of local pixels, while the deep feature map contains the local semantic information in the image. In the decoding process of JEUNet, the outputs of the 2nd, 3rd and 4th stages are input to the designed JEU (Joint Excitation Up-sampling block) module, which replaces the continuous upsampling of convolutions at three scales in the original UNet. The output of the JEU module is upsampled once and cascaded with the output of the first stage of the encoder, and then two 3×3 convolution operations are performed in the same layer. Finally, it is upsampled to the original image size and connected with the initial convolution result of the original image. After two convolution operations, the prediction result is obtained. In the decoder part, the upsampled feature map is cascaded with the shallow features of the encoder part to improve the local prediction accuracy of the pixel.

[0039] The Jointly Excited Upsampling (JEU) module is a computational unit that integrates spatial attention into the joint upsampling network. On the one hand, three feature maps with different receptive fields are resized to the same size and connected as a joint feature map T; on the other hand, a transformation vector W is constructed to excite each channel in the tensor T. During the upsampling process, the JEU module simultaneously considers the correlation of multi-scale feature information and channel direction, improving the accuracy of semantic segmentation while reducing the complexity of upsampling calculations. Figure 2 The structural diagram of the JEU block is shown, with the purple arrows indicating the joint upsampling path and the red arrows indicating the feature map excitation path.

[0040] The JEUNet of the present invention uses joint upsampling to replace the initial two layer-by-layer upsampling in the decoder link of UNet. Through embedding vector extraction, the multi-level feature maps output by the 2nd, 3rd and 4th stages are respectively input into the designed JEU module, and the feature maps of stage 4 (w×w) and stage 3 (2w×2w) are upsampled to 4w×4w of the same size as stage 2. The feature maps after joint upsampling are concat-operated to obtain a multi-level feature map T of size 4w×4w×X (X=4k+8k+16k=28k=896). After joint upsampling, the new three feature maps are concatenated and merged into a multi-level feature map T of size 4w×4w×X, where X=4k+8k+16k=28k=896. The multi-level feature map T contains 4w×4w embedding vectors of size 1×X, each of which contains semantic information of three different receptive fields and corresponds one-to-one to each position (x, y) in the original image.

[0041] After joint upsampling, the JPU block then uses 4 dilated convolutions with different expansion rates to extract features and concatenate them. The purpose is to capture feature information at different scales in the multi-level feature map to ensure the accuracy of the semantic segmentation upsampling decoder. However, each scale of the feature map in the multi-level feature map itself contains semantic information of different receptive fields. The use of multiple dilated convolutions may be repeated with the multi-level embedding vector itself, resulting in no significant improvement in the accuracy of semantic segmentation. In addition, the multi-level feature map T contains X channels, each of which is obtained by convolution kernels with different parameters. The correlation between these channels has not been paid attention to, so the accuracy improvement of the semantic segmentation prediction results is limited.

[0042] In an optional example of this implementation, for a multi-level feature map, a global average pooling function is used to compress the feature map of each channel into a feature vector. By using a global average pooling function to compress each channel in a multi-level feature map into a feature vector, efficient compression and feature extraction of the feature map are achieved, which not only reduces the dimension of the data, but also retains the key information in the channel, so that the model can more centrally process the feature response of each channel, while enhancing the model's ability to capture the global information of the feature map, and improving the model's recognition accuracy of the cigarette package ash crack feature.

[0043] In an optional example of this implementation, the feature vector is mapped through two fully connected layers to form a final weight vector, and a new response weight is assigned to each channel in the multi-level feature map obtained by joint upsampling through the weight vector, and finally a feature map after spatial channel excitation is obtained. The feature vector is mapped through two fully connected layers to form a final weight vector, and these weights are used to assign new response weights to each channel in the multi-level feature map obtained by joint upsampling, so that the model can adaptively adjust the feature response of each channel, and enhance the recognition ability of the crack area, especially in complex backgrounds and various crack forms. Through the feature map after spatial channel excitation, the model can more accurately capture the detailed information of the crack.

[0044] Specifically, in order to adaptively recalibrate the feature response in the channel direction and model the interdependence between different channels. The SE (Squeeze-and-Excitation) module is introduced in the JEU block to integrate spatial attention into the structure of the upsampling network, which is also a significant difference between the JEU and JPU modules. As shown in Formula 1, for a 4w×4w×X Tensor T, a global average pooling function is used to compress it into a 1×X feature vector F, and the xth element F in F x The calculation method is as follows:

[0045]

[0046] In the formula, F(x) corresponds to the x-th channel in Tensor T, and I and J are the x-th channel feature map T respectively. x The length and width of (I = J = 4w), i and j represent the coordinates of each point on the feature map. In order to establish the correlation between the features of different channels, F is connected to two fully connected layers to calculate the contribution weights of different channels during the learning process and to activate the corresponding channels in the feature map during prediction, as shown in formula (2).

[0047]

[0048] Where δ() represents the ReLU activation function and σ() represents the Sigmoid activation function.

[0049] During the learning process, the feature vector F is mapped by two fully connected layers to form the final weight vector W. The weight Wx of each channel in W predicts the importance of the corresponding channel x in the multi-level feature map T, thereby modeling the correlation between feature channels. In the two fully connected layer operations, the vector F is scaled (1 / 2) and restored once to reduce the amount of calculation. w1 is the mapping weight vector of the first fully connected layer, and w2 is the mapping weight vector of the second fully connected layer. Represents the excitation feature map of the feature map Tx corresponding to the xth channel and the weight vector Wx. The JEU block assigns a new response weight to each channel c in the multi-level feature map T obtained by joint upsampling through the weight vector W, and finally obtains the multi-level feature map after spatial channel excitation.

[0050] In an optional embodiment of the present invention, the joint excitation upsampling network calculation method also includes: constructing a cigarette ash crack segmentation model based on the feature map, optimizing the cigarette ash crack segmentation model using the Binary Cross-Entropy loss function, and using 10-fold Cross Validation to train the model. By using the Binary Cross-Entropy loss function optimization and 10-fold Cross Validation training method, not only the accuracy of the model in identifying the crack area is improved, but also the generalization ability of the model on different crack feature images is enhanced through cross-validation, ensuring the stability and reliability of the model in practical applications. In addition, this optimization and verification strategy also helps to discover and solve the problem of model overfitting, further improving the performance of the model on unknown data.

[0051] The present invention also provides a detection system for cigarette ash cracks, wherein the detection system implements the above method, including:

[0052] An image acquisition module, used for real-time acquisition of crack image data during cigarette burning;

[0053] A data preprocessing module is used to remove noise, calibrate grayscale and enhance contrast of image data;

[0054] The segmentation module includes the JEUNet encoder-decoder structure and the joint excitation upsampling module, which combines the upsampling operation of the multi-level feature map to achieve accurate segmentation of the crack area;

[0055] The output module is used to generate visual segmentation results and analysis reports of the crack area.

[0056] The present invention also provides a computer-readable storage medium on which a computer program is stored, wherein the computer program implements the steps of the above-mentioned calculation method when executed by a processor.

[0057] The present invention further provides a computer device, comprising a memory and a processor, wherein a computer program that can be run on the processor is stored in the memory, wherein the processor implements the steps of the above-mentioned calculation method when executing the computer program.

[0058] The specific implementation process of the present invention is now described in detail with reference to the embodiments:

[0059] The model training computing platform of the present invention is set up as follows: JEUNet is implemented on the pytorch library, and the image reading and adjustment parts call some functions in Opencv 5.0. The CPU is Core(R)i7-8700K CPU@3.70GHz×12 (Intel Corporation, USA), graphics card is GeForce RTX 3080 (Nvidia Corporation, USA). The Python version of the deep learning environment is 3.6, PyTorch version is 1.10.1, OpenCV version is 4.6, CUDA version is 11.3, and CUDNN version is 8.2. On the constructed cigarette package ash crack dataset, JEUNet randomly selects 90 images with ground truth and divides them into training set and validation set according to 10-fold Cross Validation for training, and tests on the remaining 10 images with ground truth. The images input to the network are uniformly resized to 448×448 pixels. The SGD optimizer is used during training, with momentum of 0.9 and weight_decay of 0.0001. The Binary Cross-Entropy loss function is used, the learning rate is set to 0.001, the batch_size is set to 1, and the training epochs is 100 rounds.

[0060] The present invention also continued the JEUNet optimization experiment. The JEUNet encode part was defined as Stage1-Stage4 according to the size of the feature map. In order to obtain the best algorithm combination, the paper studied the selection of feature map combinations of different stages during upsampling and input them into the JEU block and combined with the feature map excitation path for joint upsampling to obtain different semantic segmentation models. Each model was trained for 100 epochs on 90 panoramic gray images, and then tested on 10 panoramic gray images. The results of the optimization experiment are shown in Table 1. It can be seen that the best upsampling combination is Stage 4+Stage 3+Stage 2, and the IOU and AUROC reached 57.89% and 81.30% respectively. In addition, after combining the excitation path, the IOU and AUROC of JEUNet were increased to 65.73% and 88.07% respectively. The average increase ratio of nearly 8% shows that the channel attention mechanism of the JEU block plays a very important role in the upsampling process.

[0061] Table 1:

[0062]

[0063] like Figure 3 As shown in the figure, in order to evaluate the performance of CSegNet and JEUNet, a comparative experiment was conducted on the cigarette ash crack image dataset with classic models such as Otsu, FCN, FastFCN, U-Net, U-Net++ and DeepLab V3+. Each method was trained on 90 images and tested on 10 images. The Otsu method is an automatic threshold segmentation technique that selects the threshold by maximizing the variance between the foreground and the background. In the case of crack segmentation, the Otsu method can distinguish between cracks (foreground) and non-crack areas (background). FCN is a pioneering work in the field of semantic segmentation of deep learning. It uses convolutional layers instead of the fully connected layers behind traditional CNN to output heat maps and uses upsampling to restore the image size. FastFCN and the classic joint pyramid upsampling block (JPU) greatly improve the upsampling performance of FCN. U-Net, U-Net++ and DeepLabV3+ are Encoder-Decoder architectures with symmetric structures.

[0064] For testing on 10 randomly selected images from the constructed dataset evaluated by IoU, Dice, and AUROC metrics, the Ostu method was used directly, while the other models were trained on 90 ground-truth images, with each model trained 50 to 100 times to achieve convergence. In addition, a complexity analysis of the deep learning based methods was performed, including measurements of the total test time Times (in seconds (s)), the number of floating point operations per second (FLOPs, Floating Point Operations, in gigabytes (G)), and the number of parameters Params (in millions (M)). For all previous deep learning methods and the proposed method, the measured tensor input size was 10×448×448. The comprehensive evaluation results of different methods are shown in Tables 2 and Figure 4 As shown, the following conclusions can be drawn:

[0065] The proposed CSegNet and JEUNet outperform the comparison methods in terms of crack segmentation performance, among which CSegNet achieves the best pixel-level prediction accuracy (IOU and AUROC reach 66.01% and 89.16% respectively), followed by JEUNet (IOU and AUROC reach 65.73% and 88.07% respectively);

[0066] Compared with FastFCN, JEUNet's IoU and AUROC indicators are improved by nearly 30% and 15% respectively, confirming the view that "modeling the interdependencies between different channels can further improve the performance of semantic segmentation networks";

[0067] Compared with the advanced DeepLabV3+ network, CSegNet and JEUNet perform better on the dataset constructed in this paper, with both IoU and AUROC increasing by about 1% to 2%;

[0068] Although CSegNet has achieved the best detection accuracy, it has the highest computational consumption, with Times, FLOPS, and Params all higher than other methods, and is not cost-effective;

[0069] While achieving detection results similar to those of CSegNet, JEUNet has a computational parameter count of 5.93, which is 1 / 6 of DeepLabV3+ and 1 / 7 of CSegNet. It has the shortest training and testing time, and the lowest computational complexity and time cost among all algorithms. From a cost-effectiveness analysis perspective, it is the most suitable method for detecting cracks in ash packaging.

[0070] Table 2:

[0071]

[0072]

[0073] like Figure 4 As shown in Figure 2, some challenging segmentation results of different methods are shown compared with manual annotation. It can be observed that the elongated burning lines extracted by CSegNet and JEUNet have the best continuity, and the segmentation results of the cracks are also closest to the ground-truth.

[0074] The detailed explanation of the above-mentioned embodiments is only intended to explain the present invention so as to facilitate a better understanding of the present invention. However, these descriptions cannot be interpreted as limitations on the present invention for any reason. In particular, the various features described in different embodiments may also be arbitrarily combined with each other to form other embodiments. Unless there is a clear description to the contrary, these features should be understood to be applicable to any embodiment and are not limited to the described embodiments.

Claims

1. A joint excitation upsampling network calculation method for cigarette ash crack segmentation, characterized in that: The joint excitation upsampling network calculation method comprises: Acquiring a crack image during the burning process of a cigarette, and preprocessing the crack image; The encoder is used to perform multi-level feature extraction on the preprocessed crack image. Feature maps of different scales are obtained through four stages of convolution and pooling operations. Downsampling is performed between each two stages through 2×2 maximum pooling. A joint excitation upsampling module is introduced in the decoder, which includes a spatial attention mechanism and a joint excitation channel mechanism. The joint excitation upsampling module uses multi-level feature maps to jointly upsample, upsamples feature maps of different scales to the same size and performs cascade processing to generate feature maps containing multi-scale information.

2. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 1, characterized in that: The spatial attention mechanism processes the feature map through maximum pooling and average pooling operations.

3. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 1, characterized in that: The joint excitation channel mechanism generates a channel excitation vector through global average pooling and adjusts the weights of feature responses of different channels.

4. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 1, characterized in that: Utilize multi-level feature map joint upsampling, including: The sampling is upsampled by a factor of 2 through a 2×2 deconvolution operation, and each upsampled feature map is cascaded to generate a multi-level feature map with different receptive fields to improve the multi-scale perception capability of the segmentation model; after the multi-level feature maps are jointly upsampled, the feature map size is adjusted to the original image size through a 1×1 convolution.

5. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 1, characterized in that: A channel excitation path is embedded in the multi-level feature maps of the second, third and fourth stages of the decoding stage. The multi-scale crack features are gradually refined through the channel excitation path, and the dependency relationship between multiple channel features is modeled.

6. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 4, characterized in that: For multi-level feature maps, the global average pooling function is used to compress the feature map of each channel into a feature vector.

7. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 4, characterized in that: The feature vector is mapped through two fully connected layers to form the final weight vector. The weight vector is used to assign a new response weight to each channel in the multi-level feature map obtained by joint upsampling, and finally a feature map after spatial channel excitation is obtained.

8. The joint excitation upsampling network computing method for cigarette ash crack segmentation according to claim 1, characterized in that: The joint excitation upsampling network calculation method also includes: The cigarette ash crack segmentation model was constructed based on the feature map, the Binary Cross-Entropy loss function was used to optimize the cigarette ash crack segmentation model, and the 10-fold Cross Validation was used for model training.

9. A detection system for cigarette ash cracks, characterized in that: The detection system implements the method described in any one of claims 1 to 8, including: An image acquisition module, used for real-time acquisition of crack image data during cigarette burning; A data preprocessing module is used to remove noise, calibrate grayscale and enhance contrast of image data; The segmentation module includes the JEUNet encoder-decoder structure and the joint excitation upsampling module, which combines the upsampling operation of the multi-level feature map to achieve accurate segmentation of the crack area; The output module is used to generate visual segmentation results and analysis reports of the crack area.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the computing method according to any one of claims 1 to 8 are implemented.

11. A computer device comprising a memory and a processor, wherein a computer program capable of running on the processor is stored in the memory, characterized in that: When the processor executes the computer program, the steps of the computing method according to any one of claims 1 to 8 are implemented.

Images