TextDCT: Arbitrary-Shaped Text Detection via Discrete Cosine Transform Mask

TL;DR

TextDCT introduces a lightweight, anchor-free method for arbitrary-shaped scene text detection using DCT to encode masks, achieving high accuracy and efficiency across multiple datasets.

Contribution

The paper proposes a novel DCT-based mask encoding, a single-level prediction framework, and a feature awareness module for improved multi-scale text detection.

Findings

Achieves 85.1 F-measure at 17.2 FPS on CTW1500

Achieves 84.9 F-measure at 15.1 FPS on Total-Text

Demonstrates competitive performance on four challenging datasets

Abstract

Arbitrary-shaped scene text detection is a challenging task due to the variety of text changes in font, size, color, and orientation. Most existing regression based methods resort to regress the masks or contour points of text regions to model the text instances. However, regressing the complete masks requires high training complexity, and contour points are not sufficient to capture the details of highly curved texts. To tackle the above limitations, we propose a novel light-weight anchor-free text detection framework called TextDCT, which adopts the discrete cosine transform (DCT) to encode the text masks as compact vectors. Further, considering the imbalanced number of training samples among pyramid layers, we only employ a single-level head for top-down prediction. To model the multi-scale texts in a single-level head, we introduce a novel positive sampling strategy by treating the shrunk text region as positive samples, and design a feature awareness module (FAM) for spatial-awareness and scale-awareness by fusing rich contextual information and focusing on more significant features. Moreover, we propose a segmented non-maximum suppression (S-NMS) method that can filter low-quality mask regressions. Extensive experiments are conducted on four challenging datasets, which demonstrate our TextDCT obtains competitive performance on both accuracy and efficiency. Specifically, TextDCT achieves F-measure of 85.1 at 17.2 frames per second (FPS) and F-measure of 84.9 at 15.1 FPS for CTW1500 and Total-Text datasets, respectively.