Research on Automatic Recognition Algorithm of Case Images Based on Convolutional Neural Network
Yue Li, Kun-Ying Li, Qin Li · 2024
This paper proposes a novel multi-modal fusion framework for automatic classification and key information extraction of legal case documents and scene photos. Firstly, the powerful image feature extraction ability of Convolutional Neural Network (CNN) was used to extract the visual features from the case pictures. Subsequently, Transformer models are introduced to process case text information to capture long-distance dependencies and contextual relationships. In order to integrate image and text information and further improve the recognition accuracy, we use XGBoost algorithm to build a classifier based on the output features of CNN and Transformer. With its high efficiency and powerful learning ability, it can effectively deal with high-dimensional feature space. The method accurately classifies the case images and achieves superior performance in the task of automatic classification of case images, which provides a new technical path for judicial image processing.