Hierarchical Modeling Approach to Fast and Accurate Table Recognition
This work addresses the challenge of efficient table recognition for intelligent information retrieval, representing an incremental improvement over existing methods.
The paper tackles the problem of slow inference in table recognition by introducing a novel multi-task model with non-causal attention and a parallel inference algorithm, achieving faster and accurate results on two large public datasets.
The extraction and use of diverse knowledge from numerous documents is a pressing challenge in intelligent information retrieval. Documents contain elements that require different recognition methods. Table recognition typically consists of three subtasks, namely table structure, cell position and cell content recognition. Recent models have achieved excellent recognition with a combination of multi-task learning, local attention, and mutual learning. However, their effectiveness has not been fully explained, and they require a long period of time for inference. This paper presents a novel multi-task model that utilizes non-causal attention to capture the entire table structure, and a parallel inference algorithm for faster cell content inference. The superiority is demonstrated both visually and statistically on two large public datasets.