Search papers, labs, and topics across Lattice.
This paper introduces DualTSR, a unified framework for scene text image super-resolution (STISR) that integrates continuous and discrete generation processes to enhance both visual quality and character semantics. By employing conditional flow matching for image latents and absorbing-state discrete diffusion for text tokens, DualTSR eliminates the need for external OCR priors during inference, thus reducing error propagation and improving efficiency. Experimental results on the CTR-TSR dataset demonstrate that DualTSR outperforms existing methods in key metrics while significantly decreasing model size and inference time.
DualTSR achieves a remarkable 12.78 percentage point improvement in accuracy while slashing model parameters from 1.23B to just 203M.
Scene text image super-resolution (STISR) aims to recover visually plausible appearance while preserving character semantics from degraded inputs. Existing STISR systems often rely on externally generated priors or separate image and text models, resulting in error propagation and costly multi-stage inference. We present DualTSR, a unified framework that formulates STISR as coupled continuous-discrete generation. Conditional flow matching restores continuous image latents, while absorbing-state discrete diffusion reconstructs text tokens. Both processes share a multimodal transformer backbone, allowing the evolving image and text states to interact throughout generation without an external OCR prior at inference. On CTR-TSR, DualTSR achieves the best FID, LPIPS, ACC, and NED among the compared methods at both X2 and X4. On an aligned RealCE subset, it obtains the best FID, ACC, and NED with competitive LPIPS. Compared with DiffTSR at X4, DualTSR improves ACC by 12.78 percentage points while reducing the parameter count from 1.23B to 203M and end-to-end latency from 13.3s to 132ms. These results establish DualTSR as an accurate and efficient method for STISR.
