Deformable Kernel Expansion Model for Efficient Arbitrary-shaped Scene Text Detection

Scene text detection is a challenging computer vision task due to the high variation in text shapes and ratios. In this work, we propose a real-time arbitrary-shaped scene text detector named Deformable Kernel Expansion (DKE), which incorporates the merits of both segmentation and contour-based detectors. DKE employs a segmentation module to segment the shrunken text region as the text kernel, then expands the kernel to obtain text boundary by regressing the vertex-wise offsets. Generating the text kernel by segmentation enables DKE to inherit the arbitrary-shaped text region modeling capability of segmentation-based detectors. Regressing the kernel with control points enables DKE to avoid the complicated pixel-level post-processing and better learn contour deformation as the contour-based detectors. Moreover, we propose an Optimal Bipartite Graph Matching Loss that measures the matching error between the predicted points and the ground truth, which efficiently minimizes the global matching distance. Extensive experiments on five standard benchmarks demonstrate that DKE achieves a good tradeoff between accuracy and efficiency in scene text detection.