An Overview of Methods of Industrial Anomaly Detection
Siqi Yang, Shuming Liu, Penghui Shang, Hui Wang · 2024
Industrial anomaly detection plays a crucial role in modern manufacturing and production processes. In order to help researchers and professionals in the manufacturing industry understand the developing research results in this field, this paper begins with an introduction of common methods of industrial anomaly detection such as neural networks, GAN, transformer with discussing the limitations in this domain. Subsequently, owing to the emergence of the fourth wave of artificial intelligence, Advanced anomaly detection techniques based on Large Language Models (LLMs) and variants are explored, including AnomalyGPT, VisionGPT and LLMAD which have influenced a lot as the cutting-edge technologies. Additionally, based on the current developing trends of the LLMs-based method, this paper analyzes challenges and deficiencies existing in practice which have the potential to lead the way in future research. Conclusively, this overview exposes the bottleneck of traditional methods and emphasizes the revolutionary impact LLMs could have on multimodal scenes of anomaly detection while putting insight into the extend of methodical practicability.