A target detection defect identification method and system based on AI vision error correction for packaging bag ligation sealing

Abstract

The present application belongs to the technical field of intelligent packaging and visual identification, and in particular to a target detection defect identification method and system for packaging bag ligation sealing based on AI visual correction. The transparent packaging bag ligation sealing process image is processed by an industrial camera, which can identify whether the ligation ribbon correctly realizes the sealing of the transparent packaging bag. First, the sealing process image is input into a target detection model based on a convolutional neural network. This model can still accurately identify whether the transparent packaging bag is correctly sealed. The identified results are stored in a video structured manner, and the results are fed back to the sealing execution mechanism to realize automatic resealing, scrapping, and subsequent processing such as manual intervention. In addition, to improve the generalization and versatility of the model, this method can realize interactive information input to modify the inference method of the target detection model, thereby affecting the inference result to adapt to different packaging bag colors, different packaging bag transparencies, different ribbon colors, and different ribbon thicknesses.

Description

Technical Field

[0001] This invention relates to the fields of intelligent packaging and visual recognition technology, specifically to a method and system for identifying defects in packaging bag sealing based on AI visual error correction. Background Technology

[0002] With the continuous improvement of automation levels in industries such as food, daily chemicals, and pharmaceuticals, transparent packaging bags are widely used in the packaging of various products due to their advantages such as low cost, good visibility, and wide applicability. On the packaging production line, the bagging and sealing processes of transparent packaging bags are key steps to ensure the product's airtightness and appearance quality. The sealing quality directly affects the product's safety, shelf life, and market qualification rate.

[0003] Currently, the sealing process for transparent packaging bags largely relies on mechanical actuators. However, in actual production, factors such as differences in packaging bag material, variations in transparency, inconsistencies in the color and thickness of the cable ties, equipment vibration, and changes in ambient lighting can easily lead to quality defects during the sealing process, including incomplete seals, missing cable ties, misaligned cable ties, or loose seals. Failure to identify these defects in a timely manner will result in substandard products flowing into subsequent processes or even the market, increasing quality risks and production costs.

[0004] Currently, the main methods for inspecting the sealing quality of transparent packaging bags include manual sampling and inspection methods based on traditional visual rules. Manual sampling suffers from high labor intensity, strong subjectivity, low inspection efficiency, and difficulty in achieving full coverage, making it unsuitable for the quality control requirements of high-speed automated production lines. Inspection methods based on traditional image processing algorithms typically rely on rules such as fixed thresholds, edge extraction, or color features for judgment. When faced with complex backgrounds, indistinct tie features, and significant variations in lighting conditions, these methods are prone to false positives and false negatives, resulting in poor stability and versatility.

[0005] In recent years, neural network-based visual inspection technology has seen some applications in industrial inspection. However, existing technologies mostly focus on the detection of general or opaque targets. For applications such as the sealing of transparent packaging bags, which have characteristics such as highly transparent backgrounds, small targets, and significant appearance differences, there are still problems with insufficient model generalization ability and limited adaptability to different packaging parameters. In addition, existing neural network-based inspection solutions usually adopt fixed inference methods. Once the production environment, packaging bag color, or cable tie specifications change, it is often necessary to retrain the model or readjust the system parameters, which is difficult to meet the needs of industrial sites for flexibility and rapid deployment.

[0006] Meanwhile, existing sealing inspection systems mostly only provide simple inspection results output and lack an effective interactive feedback mechanism with the sealing execution mechanism. This makes it difficult to achieve automatic correction, classification and processing of sealing defects and full-process status tracking, which is not conducive to closed-loop control and quality management of the production line.

[0007] Therefore, there is an urgent need for a defect identification method and system suitable for the tying and sealing process of transparent packaging bags. This system should be able to accurately identify the tying and sealing status in complex transparent scenarios, and should have good generalization ability, interactive adjustment ability, and feedback mechanism that links with the sealing execution mechanism, so as to improve the automation level and quality control capability of the packaging production line. Summary of the Invention

[0008] (a) Technical problems to be solved

[0009] To address the shortcomings of existing technologies, this invention provides a target detection defect identification method and system based on AI visual error correction for sealing packaging bags. By introducing a target detection model based on convolutional neural networks and combining it with an interactive information input mechanism, the sealing status of transparent packaging bags is accurately identified, enabling real-time detection, structured storage, and feedback control of sealing defects. This improves the automation level, detection stability, and quality controllability of the packaging production line, solving the problems mentioned in the background technology.

[0010] (II) Technical Solution

[0011] To achieve the above objectives, the present invention specifically adopts the following technical solution:

[0012] A method for defect identification of target detection based on AI visual error correction for sealing packaging bags includes the following steps:

[0013] S1. Use an industrial camera to collect images or video data of the sealing process of transparent packaging bags to achieve real-time monitoring of the sealing process;

[0014] S2. The collected sealing process images are combined with interactive information and input into the target detection model for reasoning and recognition. The target detection model is a target detection model based on convolutional neural networks, which is used to locate and identify the tie straps on the transparent packaging bag in order to determine whether the tie straps correctly seal the transparent packaging bag.

[0015] S3. Based on the inference results output by the target detection model, analyze and judge the sealing status of the transparent packaging bag to determine whether there are sealing defects such as incomplete sealing, missing tie, or abnormal tie position.

[0016] S4. Store the reasoning and recognition results in the form of video structured data, and generate sealing status information corresponding to each sealing operation to achieve traceable management of the sealing process.

[0017] S5. Feedback the sealing status information to the sealing execution mechanism or the upper control system to trigger automatic resealing, scrapping, or manual intervention based on the detection results, so as to achieve closed-loop control of sealing quality.

[0018] In step S2, the target detection model is a multi-stage target detection model based on convolutional neural networks to improve the detection accuracy of small knotted ties and facilitate the addition of interactive information. The inference process of the multi-stage target detection model includes: an image preprocessing layer, an image feature extraction layer, a linear layer, and a post-processing layer for inference results. The image preprocessing layer typically includes operations such as image cropping, image scaling, and image transformation to improve image feature discrimination and neural network recognition. The image feature extraction layer uses a convolutional neural network to extract image features related to defects for subsequent probability induction. The linear layer inductively summarizes the image features to output the probability and location of each defect. The post-processing layer further optimizes the results of the linear layer to eliminate possible false positives. The inference results of the target detection model include one or more of the following: complete sealing, incomplete sealing, no ties detected, and abnormal ties position.

[0019] Furthermore, the interactive information includes one or more of the following: packaging bag color information, packaging bag transparency parameters, cable tie color information, defect-free image, defective image, and detection frame information.

[0020] Furthermore, the interactive information is used to modify the inference method of the object detection model. This modification includes pre-inference modification, in-inference modification, and post-inference modification. Specifically, pre-inference modification refers to the operation of the image preprocessing layer in the multi-stage object detection model modifying the image transformation based on the packaging bag color information, packaging bag transparency parameters, and cable tie color information in the interactive information. This ensures that the actual input image during deployment is as close as possible to the distribution of the multi-stage object detection model's training set, thereby improving the model's generalization and robustness. In-inference modification refers to performing gradient backpropagation on the linear layer in the multi-stage object detection model based on the defect-free image, the defective image, and detection box information, thereby fine-tuning the parameters of the linear layer to achieve accurate identification of newly emerging defects. Post-inference modification refers to adjusting the parameters of the inference result post-processing layer in the multi-stage object detection model based on the defective image and detection box information.

[0021] Furthermore, the interactive information is input through a human-computer interaction interface or a host system, and affects the inference results of the target detection model without retraining the complete target detection model, thereby improving the model's adaptability to different production scenarios.

[0022] Furthermore, in step S4, the video structured data includes at least one or more of the following: sealing sequence number, sealing time information, sealing position, cable tie detection result, defect type, and corresponding image or video index information.

[0023] Furthermore, in step S5, the sealing actuator includes a ligation device, a sealing correction device, or an automated control unit that is communicatively connected to it.

[0024] The present invention also provides a target detection defect recognition system based on AI visual error correction for sealing packaging bags, including an industrial camera module, a processor, and a memory.

[0025] The industrial camera module is used to acquire image or video data of the transparent packaging bag tying and sealing process; the memory stores a computer program; when the processor executes the computer program, it is used to implement the defect identification method steps described in any of the above method schemes.

[0026] (III) Beneficial Effects

[0027] Compared with existing technologies, this invention provides a target detection defect identification method and system based on AI visual error correction for sealing packaging bags, which has the following beneficial effects:

[0028] 1. Improve the accuracy of sealing defect detection in transparent packaging bags: By using a target detection model based on convolutional neural networks, the problem of complex backgrounds and small cable ties in transparent packaging bags can be effectively overcome, thereby improving the accuracy and stability of sealing defect identification.

[0029] 2. Introduce an interactive reasoning mechanism to enhance the model's generalization ability: The model's reasoning method is dynamically adjusted through interactive information input, adapting to changes in the color and transparency of different packaging bags, as well as the color and thickness of different cable ties, without retraining the complete model.

[0030] 3. Achieve closed-loop control of detection and feedback: Feedback the detection results to the sealing actuator or the upper control system in real time to realize automatic resealing, scrapping or manual intervention, thereby improving the automation level and quality control capability of the production line.

[0031] 4. Supports structured video storage and quality traceability: The sealing inspection results are stored in a structured video format, which facilitates subsequent quality analysis, problem tracing, and production optimization.

[0032] 5. The system has a clear structure and is easy to deploy in engineering: the method and system correspond to each other, making it easy to integrate and deploy in existing packaging production lines, and has high industrial application value. Attached Figure Description

[0033] Figure 1 This is an overall flowchart of a defect identification method based on interactive target detection, applicable to the sealing of transparent packaging bags. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Example

[0036] like Figure 1 As shown, an embodiment of the present invention proposes a target detection defect identification method based on AI visual error correction for sealing packaging bags, which includes the following steps:

[0037] S1. Acquire images or video data of the transparent packaging bag tying and sealing process using an industrial camera;

[0038] S2. The collected sealing process images are combined with interactive information and input into the target detection model for reasoning and recognition. The target detection model is a target detection model based on convolutional neural networks, which is used to identify whether the tying straps are correctly used to seal the transparent packaging bag.

[0039] S3. Based on the inference results of the target detection model, determine whether the transparent packaging bag has a sealing defect;

[0040] S4. Store the reasoning and recognition results in the form of video structured data, and generate corresponding sealing status information;

[0041] S5. Feedback the sealing status information to the sealing execution mechanism or the upper control system to trigger automatic resealing, scrapping, or manual intervention.

[0042] This method processes images of the sealing process of transparent packaging bags using an industrial camera to identify whether the tying straps are correctly used to seal the bags. First, the sealing process images are input into a convolutional neural network-based object detection model. Even with high transparency and thin straps, the model can accurately identify whether the transparent packaging bags are correctly sealed. Then, the identified results are stored in a structured video format and fed back to the sealing mechanism for automatic resealing, disposal, and manual intervention. Furthermore, to improve the model's generalization and versatility, the method allows for interactive input to modify the object detection model's inference method, thereby affecting the inference results and adapting to different packaging bag colors, transparency levels, strap colors, and thicknesses.

[0043] like Figure 1 As shown, in some embodiments, the target detection model is a multi-stage target detection model based on a convolutional neural network. This model can accurately detect small tying straps in packaging bag scenarios with high transparency. The inference process of the multi-stage target detection model includes an image preprocessing layer, an image feature extraction layer, a linear layer, and an inference result post-processing layer. The image preprocessing layer performs image cropping, scaling, normalization, and transformation operations to enhance the feature differentiation between the tying straps and the packaging bag background. The image feature extraction layer extracts multi-level image features related to the tying straps and sealing defects using a convolutional neural network. The linear layer performs inductive mapping on the extracted image features, outputting the probability of the tying straps' presence, spatial location, and sealing state probability. The inference result post-processing layer filters the output results of the linear layer, eliminating low-confidence detection results and false positive results.

[0044] like Figure 1 As shown, in some embodiments, the interactive information includes one or more of the following: packaging bag color information, packaging bag transparency parameter, cable tie color information, defect-free image, defective image, and detection box information. The interactive information is input through a human-computer interaction interface or a host system to modify the inference method of the target detection model. The modification method includes at least one of pre-inference modification, in-inference modification, and post-inference modification. Pre-inference modification involves the image preprocessing layer dynamically adjusting the image transformation operator parameters based on the packaging bag color information, packaging bag transparency parameter, and cable tie color information in the interactive information. In-inference modification involves constructing a supervised loss function based on the defect-free image, defective image, and detection box information, and fine-tuning the linear layer parameters through gradient backpropagation. Post-inference modification involves dynamically adjusting the filtering threshold or filtering rules of the post-processing layer based on the defective image and detection box information.

[0045] like Figure 1As shown, in some embodiments, the inference results of the target detection model include one or more of the following: sealing complete, sealing incomplete, no cable tie detected, and cable tie position abnormal.

[0046] like Figure 1 As shown, in some embodiments, in step S4, the video structured data includes at least one or more of the following: sealing sequence number, sealing time information, sealing location, cable tie detection result, defect type, and image or video index information.

[0047] like Figure 1 As shown, in some embodiments, in step S5, the sealing actuator includes a ligation device, a sealing correction device, or an automated control unit that is communicatively connected thereto.

[0048] The present invention also provides a target detection defect recognition system based on AI visual error correction for sealing packaging bags, including an industrial camera module, a processor, and a memory.

[0049] The industrial camera module is used to acquire image or video data of the transparent packaging bag tying and sealing process; the memory stores a computer program; when the processor executes the computer program, it is used to implement the defect identification method steps described in any of the above method schemes. Specific Implementation

[0050] The defect identification method based on interactive target detection for sealing transparent packaging bags, as described in this invention, is applied to automated packaging production lines for real-time quality inspection and feedback control of the sealing process of transparent packaging bags. The specific implementation process is as follows:

[0051] (I) Image Acquisition Steps

[0052] In step S1, an industrial camera is installed at the transparent packaging bag sealing station. The industrial camera is fixedly installed at a predetermined position that can cover the sealing area and operates synchronously with the packaging production line. The industrial camera continuously captures images of the sealing process of the transparent packaging bag according to preset frame rate, gain, and other parameters, acquiring image or video data of the sealing process. The acquired data is then transmitted in real time to the back-end processing unit to achieve real-time monitoring of the sealing process.

[0053] In this embodiment, the industrial camera can be an area scan industrial camera or a line scan industrial camera, and the acquisition method can be continuous video streaming or triggered image acquisition, with no specific form limited.

[0054] (II) Target Detection and Interactive Reasoning Steps

[0055] In step S2, the collected sealing process images, combined with interactive information, are input into the target detection model for inference and recognition. The target detection model is a multi-stage target detection model based on a convolutional neural network, used to locate and identify the binding straps on the transparent packaging bag to determine whether the binding straps correctly seal the transparent packaging bag.

[0056] In this embodiment, the inference process of the multi-stage target detection model includes the following stages:

[0057] 1. Image preprocessing layer

[0058] The image preprocessing layer performs preprocessing operations on the input sealing process image. These preprocessing operations include, but are not limited to, image cropping, image scaling, image normalization, and image transformation, which are used to enhance the feature differentiation between the tying straps and the transparent packaging bag background and improve the neural network's ability to recognize image features.

[0059] Let the image of the sealing process captured by the industrial camera be represented as:

[0060]

[0061] In the formula, H and W are the image height and width, respectively, and C is the number of color channels.

[0062] The image preprocessing layer performs the following transformations on the input sealing process image:

[0063]

[0064] In the formula, T() represents the image transformation operator, including cropping, scaling, channel color transformation, etc. The parameters of the image transformation operator include the cropping position (x, y), cropping height and width (H, W), and scaling factor (...). ), Channel color transformation function parameters .

[0065] 2. Image Feature Extraction Layer

[0066] The preprocessed image is input to the image feature extraction layer, which extracts multi-level image features related to ligatures and sealing defects through a convolutional neural network, providing a feature basis for subsequent sealing status judgment.

[0067] Image after preprocessing The input is fed into the image feature extraction layer. Let the convolutional neural network's... The layer output feature map is:

[0068]

[0069] In the formula, ; and The first Layer convolution kernel parameters and biases; " indicates convolution operation; This represents a non-linear activation function.

[0070] The final high-dimensional feature representation related to ligation band and sealing defects is obtained:

[0071]

[0072] in This represents the entire feature extraction network.

[0073] 3. Linear layer

[0074] The linear layer summarizes and maps the extracted image features, and outputs the probability of the existence of the ligation band target, its spatial location, and the probability of its corresponding sealing state.

[0075] The linear layer inductively summarizes the feature representation F and outputs the prediction results of the sealing state and target position:

[0076]

[0077] In the formula, , For linear layer parameters; Output vectors for the model.

[0078] Map the output to probabilities using the Softmax or Sigmoid function:

[0079]

[0080] In the formula, This indicates the number of sealing status categories.

[0081] The final set of sealing states output by the model is as follows:

[0082]

[0083] 4. Post-processing layer for inference results

[0084] The inference result post-processing layer further processes the results output by the linear layer to remove low-confidence detection results or possible false positive results, thereby obtaining the final target detection output.

[0085] The set of detection results output by the linear layer is denoted as:

[0086]

[0087] In the formula, Indicates the first One detection box; This indicates the corresponding detection confidence level.

[0088] The post-processing layer of inference results is based on a threshold. Perform filtering:

[0089]

[0090] The inference results of the target detection model include one or more of the following: sealing complete, sealing incomplete, no cable tie detected, and cable tie position abnormal.

[0091] (III) Mechanism of Interactive Information

[0092] In this embodiment, the interactive information includes one or more of the following: packaging bag color information, packaging bag transparency parameter, cable tie color information, defect-free image, defective image, and detection box information. This interactive information is input through a human-computer interaction interface or a host system and is used to modify the inference method of the target detection model.

[0093] Specifically, the interactive information can influence the model inference process at the following stages:

[0094] 1. Modification before reasoning

[0095] In the pre-inference stage, the image preprocessing layer dynamically adjusts the image transformation strategy based on the packaging bag color information, packaging bag transparency parameters, and cable tie color information in the interaction information, so that the distribution of input images in the actual deployment environment is as close as possible to the image distribution of the model training set, thereby improving the model's generalization ability and robustness in different production scenarios.

[0096] When the interactive information is introduced, the image transformation operator parameters Based on the color information of the packaging bag Packaging bag transparency parameters and cable tie color information Dynamic adjustment:

[0097]

[0098] In the formula, To adjust the method, the main adjustment is to the channel color transformation function parameter in the image transformation operator parameters. It can be a piecewise function, an exponential function, or a logarithmic function, etc.

[0099] This allows the actual input image distribution to be as close as possible to the model training data distribution, improving the model's generalization ability under different transparency and color conditions.

[0100] 2. Modifications during reasoning

[0101] During the inference process, gradient backpropagation is performed on the linear layer in the multi-stage target detection model based on the input defect-free image, defective image, and detection box information to fine-tune the parameters of the linear layer, thereby improving the model's ability to identify newly emerging sealing defect types.

[0102] Incorporating the defect-free image, the defective image, and the detection box information, a supervised loss function is constructed as follows:

[0103]

[0104] In the formula, This indicates the loss due to the sealed state. This indicates the loss due to the cable tie position regression. These are the weighting coefficients.

[0105] Based on gradient backpropagation, only the parameters of the linear layer are updated:

[0106]

[0107] In the formula, This is the learning rate.

[0108] This process does not involve updating the parameters of the image feature extraction layer, thus enabling rapid adaptation to newly emerging sealing defects without retraining the complete object detection model.

[0109] 3. Modification after reasoning

[0110] After inference is completed, the parameters of the post-processing layer of the inference result are adjusted according to the defective image and detection box information to optimize the detection threshold or screening rules and reduce false detections and missed detections.

[0111] After introducing interactive information, the threshold is dynamically adjusted based on the defective image and the detection box information:

[0112]

[0113] The interactive modification process described above is completed without retraining the complete object detection model, thus enabling rapid adaptation to changes in parameters of different packaging bags and cable ties.

[0114] (iv) Steps for judging sealing defects and storing data

[0115] In step S3, the sealing status of the transparent packaging bag is analyzed and judged based on the inference results output by the target detection model to determine whether there are sealing defects such as incomplete sealing, missing tie, or abnormal tie position.

[0116] In step S4, the reasoning and recognition results are stored in the form of structured video data. The structured video data includes at least one or more of the following: sealing sequence number, sealing time information, sealing location, cable tie detection result, defect type, and corresponding image or video index information, thereby enabling traceable management of the sealing process.

[0117] (v) Feedback Control Steps

[0118] In step S5, the sealing status information is fed back to the sealing actuator or the upper control system. The sealing actuator includes a ligation device, a sealing correction device, or an automated control unit that is communicatively connected to it.

[0119] In this embodiment, when the detection result indicates that there is a defect in the seal, the upper control system can trigger automatic resealing, scrapping, or manual intervention based on the seal status information, thereby achieving closed-loop control of the seal quality.

[0120] This process does not involve updating the parameters of the image feature extraction layer, thus enabling rapid adaptation to newly emerging sealing defects without retraining the complete object detection model.

[0121] In summary, this invention provides a target detection defect recognition method based on AI visual error correction for sealing packaging bags, which has the following beneficial effects:

[0122] 1. Improve the accuracy of sealing defect detection in transparent packaging bags: By using a target detection model based on convolutional neural networks, the problem of complex backgrounds and small cable ties in transparent packaging bags can be effectively overcome, thereby improving the accuracy and stability of sealing defect identification.

[0123] 2. Introduce an interactive reasoning mechanism to enhance the model's generalization ability: The model's reasoning method is dynamically adjusted through interactive information input, adapting to changes in the color and transparency of different packaging bags, as well as the color and thickness of different cable ties, without retraining the complete model.

[0124] 3. Achieve closed-loop control of detection and feedback: Feedback the detection results to the sealing actuator or the upper control system in real time to realize automatic resealing, scrapping or manual intervention, thereby improving the automation level and quality control capability of the production line.

[0125] 4. Supports structured video storage and quality traceability: The sealing inspection results are stored in a structured video format, which facilitates subsequent quality analysis, problem tracing, and production optimization.

[0126] 5. The system has a clear structure and is easy to deploy in engineering: the method and system correspond to each other, making it easy to integrate and deploy in existing packaging production lines, and has high industrial application value.

[0127] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for target detection and defect recognition based on AI visual error correction for sealing packaging bags, characterized in that, Includes the following steps: S1. Acquire images or video data of the transparent packaging bag tying and sealing process using an industrial camera; S2. The collected sealing process images are combined with interactive information and input into the target detection model for reasoning and recognition. The target detection model is a target detection model based on convolutional neural networks, which is used to identify whether the tying straps are correctly used to seal the transparent packaging bag. S3. Based on the inference results of the target detection model, determine whether the transparent packaging bag has a sealing defect; S4. Store the reasoning and recognition results in the form of video structured data, and generate corresponding sealing status information; S5. Feedback the sealing status information to the sealing execution mechanism or the upper control system to trigger automatic resealing, scrapping, or manual intervention.

2. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 1, characterized in that: The target detection model is a multi-stage target detection model based on convolutional neural networks. The inference process of the multi-stage target detection model includes an image preprocessing layer, an image feature extraction layer, a linear layer, and a post-processing layer for inference results.

3. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 2, characterized in that: The image preprocessing layer is used to perform image cropping, image scaling, image normalization and image transformation operations to enhance the feature differentiation between the tying straps and the background of the packaging bag; The image feature extraction layer extracts multi-level image features related to ligation straps and sealing defects through a convolutional neural network. The linear layer performs inductive mapping on the extracted image features and outputs the probability of the existence, spatial location, and sealing state of the ligation band target. The post-processing layer for inference results filters the output results of the linear layer, eliminating low-confidence detection results and false positive results.

4. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 1, characterized in that: The interactive information includes one or more of the following: packaging bag color information, packaging bag transparency parameters, cable tie color information, defect-free image, defective image, and detection frame information.

5. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 4, characterized in that: The interactive information is input through a human-computer interaction interface or a host system and is used to modify the inference method of the target detection model. The modification method includes at least one of pre-inference modification, in-inference modification, and post-inference modification.

6. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 5, characterized in that: The inference-before-inference modification involves the image preprocessing layer dynamically adjusting the image transformation operator parameters based on the packaging bag color information, packaging bag transparency parameters, and cable tie color information in the interaction information. The inference process is modified to construct a supervised loss function based on defect-free images, defective images, and detection box information, and fine-tune the linear layer parameters through gradient backpropagation. The post-inference modification involves dynamically adjusting the filtering threshold or filtering rules of the post-processing layer of the inference results based on the defective image and detection box information.

7. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 1, characterized in that: The inference results of the target detection model include one or more of the following: sealing complete, sealing incomplete, no cable tie detected, and cable tie position abnormal.

8. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 1, characterized in that: In step S4, the video structured data includes at least one or more of the following: sealing sequence number, sealing time information, sealing location, cable tie detection result, defect type, and image or video index information.

9. The method for target detection and defect recognition based on AI visual error correction for sealing packaging bags according to claim 1, characterized in that: In step S5, the sealing actuator includes a ligation device, a sealing correction device, or an automated control unit that is communicatively connected to it.

10. A target detection and defect recognition system based on AI visual error correction for sealing packaging bags, characterized in that, include: Industrial camera modules are used to capture image or video data of the process of tying and sealing transparent packaging bags; processor; A memory, wherein a computer program is stored; When the computer program is run on the processor, it causes the processor to perform the defect identification method as described in any one of claims 1 to 9.

Images