Seed packaging bag AI intelligent anti-counterfeiting traceability management system and method

Abstract

The invention provides a seed packaging bag AI intelligent anti-fake traceability management system and method, and relates to the field of household kitchens. The AI intelligent anti-fake traceability management system for the seed packaging bag comprises an intelligent anti-fake packaging bag which is provided with a natural random physical texture layer, a dynamic optical structure layer covering the natural random physical texture layer and an encrypted identifier storing a digital identity (DID), and the microstructure of the natural random physical texture layer forms a uniqueness feature which cannot be copied. The dynamic optical structure layer generates a variable optical effect during physical interaction. Unique biological fingerprints are formed through microcosmic random textures of a packaging base material, the optical variable effect of a microlens array is overlaid, each packaging bag has the unclonability of the physical level, and in combination with dynamic comparison of a multi-mode AI engine on real-time texture features and block chain anchoring initial values and optical living body frequency analysis, the non-clonability of the packaging bag is improved. And a three-dimensional verification closed loop which is difficult to break by a counterfeiter is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing systems, and in particular to an AI-powered intelligent anti-counterfeiting and traceability management system and method for seed packaging bags. Background Art

[0002] With the development of the Internet of Things and sensor technology, it has become possible to collect seed production and transportation environment data in real time, providing basic support for traceability. Blockchain technology, with its tamper-proof characteristics, has built a trusted distributed ledger to ensure that data from production to sales is transparent and cannot be forged. AI technology uses convolutional neural networks and multimodal anti-counterfeiting models to achieve packaging bag defect detection, authenticity verification and abnormal behavior warnings.

[0003] At present, although the existing technologies have certain applications in the field of seed anti-counterfeiting and traceability, there is still room for improvement. Traditional solutions mostly rely on a single static identification, and it is difficult to achieve non-cloning at the physical anti-counterfeiting level. In addition, the depth of multi-technology integration is insufficient, and data verification lacks dynamic multimodal collaboration. In blockchain applications, traceability data storage and physical carrier verification are prone to disconnection, and anti-fraud mechanisms mostly remain in the passive response stage, and a dynamic closed-loop verification system for physical characteristics and digital identity has not yet been formed. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides an AI intelligent anti-counterfeiting and traceability management system and method for seed packaging bags, which solves the problem of lack of dynamic multimodal collaboration in data verification of the seed anti-counterfeiting and traceability system.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an AI intelligent anti-counterfeiting and traceability management system for seed packaging bags, comprising: Smart anti-counterfeiting packaging bags: These bags have a natural random physical texture layer, a dynamic optical structure layer covering it, and an encrypted identifier that stores a digital identity (DID). The microstructure of the natural random physical texture layer creates an irreproducible unique feature. The dynamic optical structure layer produces a variable optical effect during physical interaction. The encrypted identifier that stores the digital identity (DID) is associated with a seed for full-link traceability data. User interaction terminal: Configures an adaptive collection guide program to guide users to collect physical feature data step by step. The physical feature data includes macro images of the physical texture layer, tilted video of the dynamic optical structure layer, and DID data. It integrates a lightweight edge computing module to compress the raw physical feature data to generate an encrypted summary. Multimodal AI Verification Engine: Receives the encrypted summary of physical feature data and the DID uploaded by the user interaction terminal, calculates the similarity between the real-time features and the initial features in real time through a fusion decision model, analyzes the frequency of in vivo changes in the optical video, and outputs a dynamically weighted authenticity verification result and risk level label; Blockchain evidence storage network: stores the initial feature hash value of the physical texture and binds it with the DID and the full-link traceability data of the seed, hierarchically managing the traceability event hash of the seed production, processing, and logistics links. The multimodal AI verification engine synchronously calls the initial physical features stored in the blockchain for comparison, realizing dynamic closed-loop verification of the physical carrier and digital identity.

[0006] Preferably, the natural random physical texture layer is composed of microscopic fibers or biological particles of the packaging substrate, and its initial digital features are extracted through a convolutional neural network to generate a unique hash value. The hash value is written into the blockchain as a physical trust anchor, forming a two-way binding relationship with the DID.

[0007] Preferably, the dynamic optical structure layer includes a microlens array and an underlying random pixel matrix, and the parameters of the two are optimized and matched by a genetic algorithm so that moiré fringes with a spectral response are generated when the tilt angle is greater than 15°. The dynamic optical characteristics of the moiré fringes are input into the AI engine as a liveness verification factor.

[0008] Preferably, the fusion decision model of the multimodal AI verification engine performs the following associated operations: Retrieve the initial texture hash value corresponding to the DID from the blockchain; Compare the similarity of real-time texture features uploaded by users; Analyze the frequency of moiré fringes changes in dynamic optical videos; If the texture similarity is ≥85% and the moiré frequency is within the natural jitter range of 1-3 Hz, it is judged as true.

[0009] Preferably, the fusion decision model adopts dynamic weighting of the attention mechanism. When the clarity of the texture image is lower than the threshold, the weight of the optical feature is increased by 40%-60%. When the DID verification is abnormal, the blockchain traceability data is triggered for secondary verification.

[0010] Preferably, the collaborative process between the user interaction terminal and the AI engine includes: The terminal pre-extracts texture feature summaries through a lightweight MOBILENET model; Dynamically compress optical video frame rate according to network bandwidth; Only feature summaries are transmitted to the cloud, and the original data is retained locally to ensure privacy.

[0011] Preferably, the blockchain evidence storage network adopts a hierarchical storage strategy, the hash values of key production and quality inspection events are directly uploaded to the chain, non-critical data is stored in IPFS, and DID is associated through on-chain pointers. The initial hash value of the physical feature is written as an independent data block, and modification permission is denied.

[0012] Preferably, when the multimodal AI verification engine fails verification, the anti-fraud linkage mechanism will drive the AI engine to automatically analyze the counterfeit feature pattern, upload the fraud pattern characteristics and geographic location to the blockchain, generate a counterfeit risk heat map, and push early warning information and a list of involved DIDs to related manufacturers.

[0013] An AI-powered anti-counterfeiting and traceability management method for seed packaging bags includes the following steps: Step 1: Scan the natural texture of the packaging bag substrate, generate a deep feature vector and hash it, bind the hash value, dynamic optical structure parameters, and seed traceability metadata to the DID and upload it to the blockchain; Step 2: The user interaction terminal guides the collection of physical feature data, generates an encrypted summary, and transmits it to the AI engine for multimodal dynamic verification: The multimodal AI engine integrates real-time texture features, optical dynamic effects, and initial blockchain data to output a trusted verification result; Step 3: The verification results are fed back to the blockchain to update the device reputation value, and high-risk counterfeit features are automatically entered into the anti-fraud model training library.

[0014] Preferably, the specific step of multimodal dynamic verification in step 2 includes calculating the matching degree between the real-time texture and the initial feature of the blockchain by cosine similarity, and the calculation formula is: Similarity= in Texture feature vectors collected in real time for users, The initial texture feature vector stored in the blockchain; Analyze the motion vector of moiré fringes in dynamic videos using optical flow method; If the similarity is ≥ 0.85 and the motion vector change rate is within the natural human hand jitter range of 1-3Hz, the verification passes.

[0015] The present invention provides an AI-powered intelligent anti-counterfeiting and traceability management system and method for seed packaging bags. This system has the following beneficial effects: 1. This invention uses the microscopic random texture of the packaging substrate to form a unique biometric fingerprint, superimposed with the optically variable effect of the microlens array, making each packaging bag physically unclonable. Combined with a multimodal AI engine that dynamically compares real-time texture features with blockchain-anchored initial values, and optical live frequency analysis, this achieves a three-dimensional verification closed loop that is difficult for counterfeiters to break. This anti-counterfeiting architecture, which transitions from material origin to digital intelligence, overcomes the problem of traditional static identification being easily replicated in large quantities.

[0016] 2. This invention achieves the value integration of anti-counterfeiting and traceability by relying on the full-link trust mechanism driven by blockchain. The full-cycle data of seeds from production to planting is strongly bound to the digital identity through hierarchical chain-linking, ensuring that the traceability information cannot be tampered with and can be queried through. When AI verification detects counterfeit features, it automatically analyzes the attack mode and feeds back to the physical anti-counterfeiting design end, forming a continuously optimized defense positive cycle. This not only establishes a reliable seed quality barrier for farmers, but also reconstructs the security governance model of the agricultural inputs market through the active early warning function of the risk heat map. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Schematic diagram of the system flow of the present invention; Figure 2 This is a schematic diagram of a user interaction terminal data collection method of the present invention; Figure 3 This is a flow chart of the multimodal AI verification engine of the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] Example: like Figure 1-Figure 3 As shown, an embodiment of the present invention provides an AI intelligent anti-counterfeiting and traceability management system for seed packaging bags, including an intelligent anti-counterfeiting packaging bag: a natural random physical texture layer, a dynamic optical structure layer covering it, and an encrypted identifier storing a digital identity DID. The microstructure of the natural random physical texture layer forms an unreplicable unique feature, the dynamic optical structure layer produces a variable optical effect during physical interaction, and the encrypted identifier storing the digital identity DID is associated with the full-link traceability data of the seed. The natural random physical texture layer is composed of microscopic fibers or biological particles of the packaging substrate. Its initial digital features are extracted through a convolutional neural network and a unique hash value is generated. The hash value is written into the blockchain as a physical trust anchor, forming a two-way binding relationship with the DID. The dynamic optical structure layer includes a microlens array and an underlying random pixel matrix. The parameters of the two are optimized and matched by a genetic algorithm, so that a spectral response moiré fringe is generated when the tilt angle is greater than 15°. The dynamic optical features of the moiré fringe are input into the AI engine as a liveness verification factor.

[0020] User interaction terminal: Configure an adaptive collection guide program to guide users to collect physical feature data step by step. The physical feature data includes macro images of the physical texture layer, tilted videos of the dynamic optical structure layer, and DID data. It integrates a lightweight edge computing module to compress the original physical feature data to generate an encrypted summary.

[0021] Multimodal AI Verification Engine: Receives the encrypted summary of physical feature data and DID uploaded by the user interaction terminal. Using a fusion decision model, it calculates the similarity between real-time features and initial features in real time, analyzes the frequency of in vivo changes in optical video, and outputs dynamically weighted authenticity verification results and risk level labels. The fusion decision model of the multimodal AI verification engine performs the following associated operations: Retrieve the initial texture hash value corresponding to the DID from the blockchain; Compare the similarity of real-time texture features uploaded by users; Analyze the frequency of moiré fringes changes in dynamic optical videos; The fusion decision model uses an attention mechanism for dynamic weighting. When the texture image clarity is below a threshold, the optical feature weight is increased by 40%-60%. When DID verification is abnormal, a secondary verification of the blockchain traceability data is triggered. The collaborative process between the user interaction terminal and the AI engine includes: The terminal pre-extracts texture feature summaries through a lightweight MOBILENET model; Dynamically compress optical video frame rate according to network bandwidth; Only feature summaries are transmitted to the cloud, and the original data is retained locally to ensure privacy.

[0022] Blockchain evidence storage network: stores the initial feature hash value of the physical texture and binds it with the DID and the full-link traceability data of the seed. It hierarchically manages the traceability event hash of the seed production, processing, and logistics links. The multimodal AI verification engine synchronously calls the initial physical features stored in the blockchain for comparison to achieve dynamic closed-loop verification of the physical carrier and digital identity. The blockchain evidence storage network adopts a hierarchical storage strategy. The hash values of key production and quality inspection events are directly uploaded to the chain, and non-critical data is stored in IPFS. The DID is associated with the on-chain pointer. The initial hash value of the physical feature is written as an independent data block, and modification permissions are denied.

[0023] When the multimodal AI verification engine fails to verify, the anti-fraud linkage mechanism will drive the AI engine to automatically analyze the counterfeit feature pattern, upload the fraud pattern characteristics and geographic location to the blockchain, generate a counterfeit risk heat map, and push early warning information and a list of involved DIDs to related manufacturers.

[0024] An AI-powered anti-counterfeiting and traceability management method for seed packaging bags includes the following steps: Step 1: Scan the natural texture of the packaging bag substrate, generate a deep feature vector and hash it, bind the hash value, dynamic optical structure parameters, and seed traceability metadata to the DID and upload it to the blockchain; Step 2: The user interaction terminal guides the collection of physical feature data, generates an encrypted summary, and transmits it to the AI engine for multimodal dynamic verification: The multimodal AI engine integrates real-time texture features, optical dynamic effects, and initial blockchain data to output a trusted verification result; Step 3: The verification results are fed back to the blockchain to update the device reputation value, and high-risk counterfeit features are automatically entered into the anti-fraud model training library.

[0025] The specific steps of multimodal dynamic verification in step 2 include calculating the matching degree between the real-time texture and the initial features of the blockchain through cosine similarity. The calculation formula is: Similarity= in Texture feature vectors collected in real time for users, The initial texture feature vector stored in the blockchain; Analyze the motion vector of moiré fringes in dynamic videos using optical flow method; If the similarity is ≥ 0.85 and the motion vector change rate is within the natural human hand jitter range of 1-3Hz, the verification passes.

[0026] Intelligent anti-counterfeiting and traceability of hybrid rice seeds: The packaging bag is made of biodegradable hemp fiber. The natural fiber distribution on its surface is imaged by a 1200dpi industrial scanner. The 1024-dimensional feature vector FINITFINI is extracted through ResNet-50, and the SHA-256 hash value HINITHINIT is generated. The microlens array is covered with an AI-generated random pixel matrix, optimized by a genetic algorithm to ensure green spectral moiré fringes when tilted 18°; Bind HINITHINIT to DID and write the associated seed production data into the blockchain; Farmers conduct procurement verification: Farmers use a mobile phone app to scan the packaging bag, take macro photos of the hemp fiber texture, tilt the phone to shoot a 5-second video to capture the dynamic changes of the moiré pattern, and scan the code to obtain the DID; MobileNetV3 compresses texture features into a 182KB summary, adaptively reduces the video frame rate to 3fps in a weak network environment, and encrypts the video before transmitting it to the cloud. AI dynamically verifies cosine similarity calculation: Similarity ≥ 0.85 and frequency 1 Hz ≤ ≤ 3 Hz, verification passed; The frequency is compliant but the Similarity is less than 0.85, indicating suspected packaging substitution and a Level 1 risk. Similarity ≥ 0.85 but with abnormal frequency, indicating a high-level counterfeit attack, indicating a high-risk product. Double abnormality: proactively identify the source of counterfeit goods and report to users for emergency treatment; Blockchain traceability display: Farmers’ mobile phones display a visual timeline and the temperature and humidity records of the entire cold chain logistics of the products; Active anti-fraud defense: Counterfeiters imitated moiré patterns but kept the frequency at 0.5Hz in an attempt to introduce high-quality counterfeit packaging bags into the market. The system responded with a closed loop: The AI engine identifies abnormal frequencies, marks high risks, and analyzes counterfeit signature patterns; Generate a heat map of counterfeit goods and upload it to the blockchain; Push parameter optimization instructions to packaging manufacturers; The manufacturer's next-generation packaging bag optical structure is produced according to new parameters, making counterfeiting twice as difficult.

[0027] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An AI intelligent anti-counterfeiting and traceability management system for seed packaging bags, characterized by: include: Smart anti-counterfeiting packaging bags: These bags have a natural random physical texture layer, a dynamic optical structure layer covering it, and an encrypted identifier that stores a digital identity (DID). The microstructure of the natural random physical texture layer creates an irreproducible unique feature. The dynamic optical structure layer produces a variable optical effect during physical interaction. The encrypted identifier that stores the digital identity (DID) is associated with a seed for full-link traceability data. User interaction terminal: Configures an adaptive collection guide program to guide users to collect physical feature data step by step. The physical feature data includes macro images of the physical texture layer, tilted video of the dynamic optical structure layer, and DID data. It integrates a lightweight edge computing module to compress the raw physical feature data to generate an encrypted summary. Multimodal AI Verification Engine: Receives the encrypted summary of physical feature data and the DID uploaded by the user interaction terminal, calculates the similarity between the real-time features and the initial features in real time through a fusion decision model, analyzes the frequency of in vivo changes in the optical video, and outputs a dynamically weighted authenticity verification result and risk level label; Blockchain evidence storage network: stores the initial feature hash value of the physical texture and binds it with the DID and the full-link traceability data of the seed, hierarchically managing the traceability event hash of the seed production, processing, and logistics links. The multimodal AI verification engine synchronously calls the initial physical features stored in the blockchain for comparison, realizing dynamic closed-loop verification of the physical carrier and digital identity.

2. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 1 is characterized by: The natural random physical texture layer is composed of microscopic fibers or biological particles of the packaging substrate. Its initial digital features are extracted through a convolutional neural network and a unique hash value is generated. The hash value is written into the blockchain as a physical trust anchor, forming a two-way binding relationship with the DID.

3. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 1 is characterized by: The dynamic optical structure layer includes a microlens array and an underlying random pixel matrix. The parameters of the two are optimized and matched by a genetic algorithm, so that moiré fringes with spectral response are generated when the tilt angle is greater than 15°. The dynamic optical characteristics of the moiré fringes are input into the AI engine as a liveness verification factor.

4. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 1 is characterized by: The fusion decision model of the multimodal AI verification engine performs the following associated operations: Retrieve the initial texture hash value corresponding to the DID from the blockchain; Compare the similarity of real-time texture features uploaded by users; Analyze the frequency of moiré fringes changes in dynamic optical videos; If the texture similarity is ≥85% and the moiré frequency is within the natural jitter range of 1-3 Hz, it is judged as true.

5. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 4 is characterized by: The fusion decision model adopts dynamic weighting of the attention mechanism. When the clarity of the texture image is lower than the threshold, the weight of the optical feature is increased by 40%-60%. When the DID verification is abnormal, the blockchain traceability data is triggered for secondary verification.

6. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 1 is characterized by: The collaborative process between the user interaction terminal and the AI engine includes: The terminal pre-extracts texture feature summaries through a lightweight MOBILENET model; Dynamically compress optical video frame rate according to network bandwidth; Only feature summaries are transmitted to the cloud, while the original data is stored locally to ensure privacy.

7. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 1 is characterized by: The blockchain evidence storage network adopts a hierarchical storage strategy. The hash values of key production and quality inspection events are directly uploaded to the chain, and non-critical data is stored in IPFS. DID is associated with the DID through on-chain pointers. The initial hash value of the physical feature is written as an independent data block, and modification permission is denied.

8. The AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to claim 1 is characterized by: When the multimodal AI verification engine fails to verify, the anti-fraud linkage mechanism will drive the AI engine to automatically analyze the counterfeit feature pattern, upload the fraud pattern characteristics and geographic location to the blockchain, generate a counterfeit risk heat map, and push early warning information and a list of involved DIDs to related manufacturers.

9. An AI intelligent anti-counterfeiting and traceability management method for seed packaging bags, using the AI intelligent anti-counterfeiting and traceability management system for seed packaging bags according to any one of claims 1 to 8, characterized in that: The following steps are involved: Step 1: Scan the natural texture of the packaging bag substrate, generate a deep feature vector and hash it, bind the hash value, dynamic optical structure parameters, and seed traceability metadata to the DID and upload it to the blockchain; Step 2: The user interaction terminal guides the collection of physical feature data, generates an encrypted summary, and transmits it to the AI engine for multimodal dynamic verification: The multimodal AI engine integrates real-time texture features, optical dynamic effects, and initial blockchain data to output a trusted verification result; Step 3: The verification results are fed back to the blockchain to update the device reputation value, and high-risk counterfeit features are automatically entered into the anti-fraud model training library.

10. The AI intelligent anti-counterfeiting and traceability management method for seed packaging bags according to claim 9, characterized in that: The specific steps of the multimodal dynamic verification in step 2 include calculating the matching degree between the real-time texture and the initial features of the blockchain through cosine similarity, and the calculation formula is: Similarity= in Texture feature vectors collected in real time for users, The initial texture feature vector stored in the blockchain; Analyze the motion vector of moiré fringes in dynamic videos using optical flow method; If the similarity is ≥ 0.85 and the motion vector change rate is within the natural human hand jitter range of 1-3Hz, the verification passes.

Images