TPFNet: A Novel Text In-painting Transformer for Text Removal

TL;DR

TPFNet is an innovative end-to-end transformer-based network that effectively removes text from images by combining feature synthesis, segmentation, and high-pass filtering, outperforming previous methods on multiple datasets.

Contribution

Introduces TPFNet, a novel one-stage transformer-based architecture with a multi-headed decoder and segmentation guidance for precise text removal from images.

Findings

Outperforms recent methods on Oxford, SCUT, and SCUT-EnsText datasets.

Achieves higher PSNR and lower text detection precision, indicating better quality and accuracy.

Utilizes a pyramidal vision transformer and adversarial loss conditioned on segmentation for improved results.

Abstract

Text erasure from an image is helpful for various tasks such as image editing and privacy preservation. In this paper, we present TPFNet, a novel one-stage (end-toend) network for text removal from images. Our network has two parts: feature synthesis and image generation. Since noise can be more effectively removed from low-resolution images, part 1 operates on low-resolution images. The output of part 1 is a low-resolution text-free image. Part 2 uses the features learned in part 1 to predict a high-resolution text-free image. In part 1, we use "pyramidal vision transformer" (PVT) as the encoder. Further, we use a novel multi-headed decoder that generates a high-pass filtered image and a segmentation map, in addition to a text-free image. The segmentation branch helps locate the text precisely, and the high-pass branch helps in learning the image structure. To precisely locate the text, TPFNet employs an adversarial loss that is conditional on the segmentation map rather than the input image. On Oxford, SCUT, and SCUT-EnsText datasets, our network outperforms recently proposed networks on nearly all the metrics. For example, on SCUT-EnsText dataset, TPFNet has a PSNR (higher is better) of 39.0 and text-detection precision (lower is better) of 21.1, compared to the best previous technique, which has a PSNR of 32.3 and precision of 53.2. The source code can be obtained from https://github.com/CandleLabAI/TPFNet