… supervised scene text detection framework (WeText) that is capable of learning a robust and
accurate scene text detector … The idea is to first train a “light” supervised model by using a …

… We use a separately trained scene text detector and an inpainting network. The scene text
detector predicts segmentation maps of text instances which are then used as masks for the …

… text detection employing a semantic segmentation network but, since they cannot rely on a
pixel–level supervision at the text … inside the bounding–box as text. Therefore, the scarcity of …

… by exploiting word annotations in rich large-scale real scene text datasets. … weakly supervised
semantic segmentation [34, 36–38], we propose a weakly supervised scene text detector …

… networks for scene text segmentation. … supervised learning approach is used to reduce the
shift between training on real and synthetic data. Pixel–level supervisions for a text detection …

… trained with weak supervision only (ie partial texts in images without text bounding boxes).
Our pre-trained model weights can be directly transferred to various scene text detectors and …

… field of scene text detection requires complex … as weak supervision, can reduce the amount
of data needed. In this paper we propose a weak supervision method for scene text detection, …

… (1) The work is a further exploration of weak supervision methods in scene text detection.
(2) The network we designed uses a U-Net shape architecture rather than the CAM architecture…

… Our model uses the image’s paired caption as weak supervision to … scene graphs without
help from texts. To learn our model, we first allow context information to propagate on the text …

… is used for training text detection model, has … supervised scene text detection framework
(SemiText), which trains robust and accurate scene text detectors using a pre-trained supervised …