Scene Text Recognition with Semantics
This addresses a practical limitation in scene text recognition for real-world applications where text images are often imperfect.
The paper tackles the problem of scene text recognition failing on noisy or partially obscured text by using semantic information from the scene to contextualize predictions, resulting in higher performance than traditional models, especially on noisy instances.
Scene Text Recognition (STR) models have achieved high performance in recent years on benchmark datasets where text images are presented with minimal noise. Traditional STR recognition pipelines take a cropped image as sole input and attempt to identify the characters present. This infrastructure can fail in instances where the input image is noisy or the text is partially obscured. This paper proposes using semantic information from the greater scene to contextualise predictions. We generate semantic vectors using object tags and fuse this information into a transformer-based architecture. The results demonstrate that our multimodal approach yields higher performance than traditional benchmark models, particularly on noisy instances.