Regressing text in natural scenes with polygonal representations is challenging due to shape prediction difficulties. To address this, we introduce Text Polygon Detection with Split Transformers (TPD-STR), which directly regresses polygonal points. TPD-STR incorporates the Decoder Split (DS) architecture to separate polygonal point regression and textness classification, and the Positional Information Propagation (PIP) module to enhance classification. Both modules are effective and compatible with existing methods. TPD-STR achieves state-of-the-art (SOTA) performance among regression-based methods, surpassing segmentation-based methods on MSRA-TD500 without external data. Adding DS and PIP to existing models further improves performance. Experiments demonstrate the model's ability to detect text instances effectively.

In this work, we introduce a new algorithm for analyzing a diagram, which contains visual and textual information in an Abstract and integrated way. Whereas diagrams contain richer information compared with individual image-based or language-based data, proper solutions for automatically understanding them have not been proposed due to their innate characteristics of multi-modality and arbitrariness of layouts. To tackle this problem, we propose a unified diagram-parsing network for generating knowledge from diagrams based on an object detector and a recurrent neural network designed for a graphical structure. Specifically, we propose a dynamic graph-generation network that is based on dynamic memory and graph theory. We explore the dynamics of information in a diagram with activation of gates in gated recurrent unit (GRU) cells. On publicly available diagram datasets, our model demonstrates a state-of-the-art result that outperforms other baselines. Moreover, further experiments on question answering shows potentials of the proposed method for various applications.

In the past, most historical research has been manually carried out by exploring historical facts reading between the lines of documents. Nowadays, historical big data has become electronically available and advances in machine learning techniques allow us to analyze the vast amount of historical data. From a historical perspective, making inferences about political stances of historical figures is important for grasping historical rivalries and power structures of an era. Thus, in this paper, we propose an approach to the systematic inference of power mechanisms based on a human network constructed from historical data. In this network, humans are linked according to the degree of kinship using genealogy records, and identified by political stances on agendas recorded in the annals of a dynasty as a political force. And then, a machine learning algorithm, semi-supervised learning, classifies humans who cannot identify political stances as political forces that reflect the links of the networks. The data consist of the genealogy of the Andong Gwon clan, a record of family relations of 10,243 people from the 10th to 15th century Korea, and the Annals of the Joseon Dynasty, a historical volume that describes historical facts of the Joseon Dynasty for 472 years and is composed of 1894 fascicles and 888 books. From the data, we construct a human network based on a historically meaningful period (1443–1488), and classify people into two political forces using the proposed method. We suggest that this machine learning approach to historical study could be utilized as a potent reference tool devoid of the subjectivism of human experts in the field of history.

In this paper, we propose an effective online method to recognize handwritten music symbols. Based on the fact that most music symbols can be regarded as combinations of several basic strokes, the proposed method first classifies all the strokes comprising an input symbol and then recognizes the symbol based on the results of stroke classification. For stroke classification, we propose to use three types of features, which are the size information, the histogram of directional movement angles, and the histogram of undirected movement angles. When combining classified strokes into a music symbol, we utilize their sizes and spatial relation together with their combination. The proposed method is evaluated using two datasets including HOMUS, one of the largest music symbol datasets. As a result, it achieves a significant improvements of about 10% in recognition rates compared to the state-of-the-art method for the datasets. This shows the superiority of the proposed method in online handwritten music symbol recognition.