Text-based person search (TBPS) is of significant importance in intelligent surveillance, which aims to retrieve pedestrian images with high semantic relevance to a given text description. This retrieval task is characterized with both modal heterogeneity and fine-grained matching. To implement this task, one needs to extract multi-scale features from both image and text domains, and then perform the cross-modal alignment. However, most existing approaches only consider the alignment confined at their individual scales, e.g., an image-sentence or a region-phrase scale. Such a strategy adopts the presumable alignment in feature extraction, while overlooking the cross-scale alignment, e.g., image-phrase. In this paper, we present a transformer-based model to extract multi-scale representations, and perform Asymmetric Cross-Scale Alignment (ACSA) to precisely align the two modalities. Specifically, ACSA consists of a global-level alignment module and an asymmetric cross-attention module, where the former aligns an image and texts on a global scale, and the latter applies the cross-attention mechanism to dynamically align the cross-modal entities in region/image-phrase scales. Extensive experiments on two benchmark datasets CUHK-PEDES and RSTPReid demonstrate the effectiveness of our approach.

Abstract Patients always want a precise diagnosis and appropriate treatment advice when they are diagnosed with a disease. Furthermore, the original information about certain features and symptoms appears in different forms, and the effect of time is always ignored in the process of diagnosis. To overcome these defects, this paper develops a systematic multiperiod hybrid decision support model, which combines the similarity measurement and three‐way decision theory to provide prediction and treatment advice for patients under the fuzzy environment. This multiperiod hybrid decision support model, which considers the effect of time, including transformation module, multiperiod integration module, similarity module, prediction module, and three‐way decision module, provides disease prediction and advice on treatment based on similarities and three‐way decision theory. To validate this model, we construct an illustration composed of four cases, and this ultimately shows that MPH‐SDM can effectively support patients' disease diagnoses and treatment.