Attention-based Extraction of Structured Information from Street View Imagery
This addresses the challenge of automated text extraction from real-world images for applications like mapping and data analysis, representing a strong specific gain rather than a foundational breakthrough.
The paper tackles the problem of extracting structured information like street names and business names from street view imagery, achieving 84.2% accuracy on the FSNS dataset, which significantly outperforms the previous state-of-the-art of 72.46%.
We present a neural network model - based on CNNs, RNNs and a novel attention mechanism - which achieves 84.2% accuracy on the challenging French Street Name Signs (FSNS) dataset, significantly outperforming the previous state of the art (Smith'16), which achieved 72.46%. Furthermore, our new method is much simpler and more general than the previous approach. To demonstrate the generality of our model, we show that it also performs well on an even more challenging dataset derived from Google Street View, in which the goal is to extract business names from store fronts. Finally, we study the speed/accuracy tradeoff that results from using CNN feature extractors of different depths. Surprisingly, we find that deeper is not always better (in terms of accuracy, as well as speed). Our resulting model is simple, accurate and fast, allowing it to be used at scale on a variety of challenging real-world text extraction problems.