Poster: Automatic Detection of Minimal Repeated Patterns in Printed Fabrics of Ethnic Minority Costumes

Printed fabrics in ethnic minority costumes feature rich and complex patterns that carry significant cultural and historical meanings. Detecting minimal repeated patterns (MRPs) in these fabrics is crucial for preserving cultural heritage and enabling intelligent production. This paper proposes an automatic MRP detection method tailored for printed fabrics of ethnic minority costumes. Building upon the Adjacent Key-points Localization (AKL) framework, we introduce innovations such as adaptive window size selection and deep learning feature extraction, naming it DeepAKL. To address the lack of public datasets, we constructed a dataset of approximately 8,000 images encompassing diverse patterns and styles to test the robustness of our algorithm. Experimental results demonstrate that our method outperforms existing techniques in accuracy and robustness, effectively detecting MRPs in complex ethnic patterns.