… a feature distillation … text encoder by leveraging it to distill the recognition decoder. As the
input and output of them are symmetrically related, we let the deep layers of the recognition …

… efforts: Scene Text Recognition (STR) that deals with text in everyday … featuremap. In order
to circumvent the discrepancy due to variable character-width, we perform feature distillation …

… This paper will focus on research of scene text detection. … with knowledge distillation for scene
text detection. There are … task loss function is designed both for feature regression and text …

… character-level, named Character-toCharacter Distillation (CCD), as shown in Fig.2, which
enables feature … Synthetic data and artificial neural networks for natural scene text recognition. …

… text location. Furthermore, we apply self-distillation to train the ISNet and refine its detection
… , ICDAR2017 MLT, CTW1500 and Total-Text, and the experimental results show that it can …

… dual features with each other, then we explore the dual distillation with a co-teacher framework.
Our proposed method shows the effectiveness of self-supervised scene text recognition …

… a new knowledge distillation method for text detection ,which … on the text of the teacher
network is distilled. Migrating the … work of scene text detection and knowledge distillation ,then …

… distillation method for highly accurate scene text recognition. The key advance of our method
is to enhance the text recognition … , which allows the text recognition model to consider both …

… knowledge distillation method via decoupled target for training accurate compact scene
text detection … This paper proposes a distillation method via decoupled target and wavelet …

… To reduce inference time and maintain a certain level of performance enhancement, we
propose a recursive distillation strategy. Specifically, the strategy first pre-trains the model with a …