Currently, deep learning-based visual inspection has been highly successful with the help of supervised learning methods. However, in real industrial scenarios, the scarcity of defect samples, the cost of annotation, and the lack of $a$ priori knowledge of defects may render supervised-based methods ineffective. In recent years, unsupervised anomaly localization (AL) algorithms have become more widely used in industrial inspection tasks. This article aims to help researchers in this field by comprehensively surveying recent achievements in unsupervised AL in industrial images using deep learning. The survey reviews more than 120 significant publications covering different aspects of AL, mainly covering various concepts, challenges, taxonomies, benchmark datasets, and quantitative performance comparisons of the methods reviewed. In reviewing the achievements to date, this article provides detailed predictions and analysis of several future research directions. This review provides detailed technical information for researchers interested in industrial AL and who wish to apply it to the localization of anomalies in other fields.

In recent years, various food-safety issues have aroused public concern regarding safety in the food supply chain. Since grains are closely linked to human life and health, it is necessary to effectively manage information in the grain supply chain. The grain supply chain is characterized by a long life cycle, complex links, various hazards, and heterogeneous information sources. Problems with traditional traceability systems include easy data tampering, difficult hazardous-material information management, the “information isolated island” problem, and low traceability efficiency in the whole supply chain. Blockchain is a distributed computing paradigm characterized by decentralization, network-wide recording, security, and reliability. As such, it can reduce administrative costs and improve the efficiency of information management. Based on literature research and a field investigation of wheat-processing enterprises in Shandong Province, We analyze the operation process of grain supply chain. This study, therefore, proposed a new system architecture in the entire grain supply chain based on blockchain technology and designed a multimode storage mechanism that combines chain storage. This prototype system was tested and verified using actual cases and application scenarios. Compared to traditional systems, the proposed system is characterized by data security and reliability, information interconnection and intercommunication, real-time sharing of hazardous-material information, and dynamic and credible whole-process tracing. As such, this system is highly significant and has reference value for guaranteeing food quality and safety-process traceability.

With the development of advanced information and intelligence technologies, precision agriculture has become an effective solution to monitor and prevent crop pests and diseases. However, pest and disease recognition in precision agriculture applications is essentially the fine-grained image classification task, which aims to learn effective discriminative features that can identify the subtle differences among similar visual samples. It is still challenging to solve for existing standard models troubled by oversized parameters and low accuracy performance. Therefore, in this paper, we propose a feature-enhanced attention neural network (Fe-Net) to handle the fine-grained image recognition of crop pests and diseases in innovative agronomy practices. This model is established based on an improved CSP-stage backbone network, which offers massive channel-shuffled features in various dimensions and sizes. Then, a spatial feature-enhanced attention module is added to exploit the spatial interrelationship between different semantic regions. Finally, the proposed Fe-Net employs a higher-order pooling module to mine more highly representative features by computing the square root of the covariance matrix of elements. The whole architecture is efficiently trained in an end-to-end way without additional manipulation. With comparative experiments on the CropDP-181 Dataset, the proposed Fe-Net achieves Top-1 Accuracy up to 85.29% with an average recognition time of only 71 ms, outperforming other existing methods. More experimental evidence demonstrates that our approach obtains a balance between the model’s performance and parameters, which is suitable for its practical deployment in precision agriculture art applications.

Aiming at the problems such as slow traceability efficiency, poor sharing, and the difficulty of matching the throughput of a blockchain single chain structure due to the complexity of the grain food supply chain links, the large number of participants, and the large amount of data information, this paper proposes a grain food blockchain traceability information management model based on the master-slave multichain structure by analyzing the processes and data characteristics of each link in the grain food supply chain; on this basis, the PLEW consensus algorithm based on Raft + improved PoW algorithm is designed for the master chain, and the CI-PBFT consensus algorithm based on trusted information degree is designed for the slave chain. The master chain and slave chain are anchored to each other through hash locking, and the data is uploaded and queried through smart contracts. In order to verify the effectiveness of the model, the blockchain traceability system is designed and implemented based on Hyperledger Fabric2.2. At the same time, it is compared with the transaction throughput and traceability efficiency of the blockchain single chain structure. Through the safety analysis of the data information of a company in Hubei, the results show that the grain traceability system designed and implemented in this study has certain advantages over the blockchain single chain structure in all aspects. It can also solve the grain food security problems that consumers worry about, and provide reference for the research of grain blockchain traceability information management.

Rice is one of the three major staple foods in the world, and the quality and safety of rice are related to the development of human beings. The new crown epidemic, pesticide residues, insect pests, and heavy metal pollution have a certain security impact on the food supply chain. The rice supply chain is characterized by a long life cycle; complex roles in the main links; many types of hazards; and multidimensional, multisource, and heterogeneous information. To strengthen the rice supply chain’s supervision ability under the epidemic situation, a supervision cross-chain model suitable for the complicated data of the rice supply chain based on parallel blockchain theory and smart contract technology was built. Firstly, the data collected in the rice supply chain and different types of data stored in different parallel blockchains were analyzed. Secondly, based on data analysis, a collection/supervision cross-chain mechanism based on “hash lock + smart contract + relay chain”, a concurrency mechanism based on the K-means algorithm and a Bloom filter, and a consensus mechanism suitable for multichain consensus named the Supervision Practical Byzantine Fault Tolerance (SPBFT) were proposed. Furthermore, a cross-chain model of rice supply chain supervision was constructed. Finally, theoretical verification and simulation experiments were used to analyze the operation process, safety, cross-chain efficiency, and scalability of the model. The results showed that the application of parallel blockchains and smart contracts to supervision of rice supply chain information improved the convenience and security of information interaction between various links in the rice supply chain, the storage cost of supply chain data and the high latency of interaction was reduced, and the refined management of the rice supply chain data and personnel was realized. This research applied new information technology to the coordination and resource sharing of the food supply chain, and provides ideas for the digital transformation of the food industry.

With the development of Agriculture 4.0, the requirements for sustainable agriculture and precision agriculture continue to grow. As one of the three major staple foods globally, the quality and safety of rice affect human health as well as social development. To ensure the quality and safety of rice and reduce the flow of problematic rice, a multi-layer blockchain-based rice supply chain refinement supervision model MBRRSM (Multi-blockchain Rice Refined Supervision Model) is proposed from the information level. First, the characteristics of information flow in the rice supply chain are analyzed, and a classification table of key information is constructed. Second, the MBRRSM framework is designed. Based on a multi-party hybrid encryption algorithm, secure multi-party computing algorithm, multi-mode storage mechanism, and SPOP (Supervision Proof of Peers) consensus algorithm, a set of mechanisms is designed for the transmission, use, storage, and consensus of rice data in MBRRSM. Subsequently, the security and performance capabilities of MBRRSM are analyzed. Meanwhile, the SPOP consensus algorithm is analyzed. Finally, a prototype system is built based on MBRRSM and verified through exemplary scenarios in different usage situations. The results show that research on the refined supervision of the rice supply chain based on multi-blockchain can finely supervise all types of data in the rice supply chain, and provide a guarantee for enterprise users to safely transmit and use data with different privacy levels. This study presents a unique research paradigm that introduces the theories and methods of the new information field generation into the field of agricultural research, and thus assists in the strategy implementation of “holding grain in the land and storing grain in technology”.