Search

Low illumination, under which discriminative clues are buried in the captured images, is an under-investigated but noteworthy issue in wild scene text detection. Existing deep learning approaches suffer from the scarcity of training data and illumination-sensitive feature representation. To address these issues, we propose a Low Illumination Scene Text (LIST) Detector training with authentic synthetic data and integrating dedicated feature enhancement modules. Specifically, we adopt a lightweight and non-reference low-light scene text image synthesis network to acquire adequate training data through pixel-wisely adjusting the dynamic range curve of normal-light images. Moreover, illumination invariant feature representation is learned through dual path feature extraction stem with intensity adjusted inputs, and feature fusion branch with automatically designed fusion cell. In the end, the enhanced feature is fed into a segmentation-based layer to localize arbitrary shape text instances. We construct a labeled real-world scene text image dataset called "DarkText" and conduct extensive experiments to validate the advantages of our proposed framework over state-of-the-art competitors. [ABSTRACT FROM AUTHOR]

• A realistic and diversified training dataset (POL) is constructed by optical modeling for reflection obstructed image. • A progressive polarization based reflection removal network (P 2 R 2 Net) with up-to-date neural network units is proposed. • The improvement of the proposed method for real-world images is verified by comparisons with the state-of-the-art methods. • Robustness analysis on reflection-tainted synthesis images suggests our method possesses excellent generalization ability. The reflection effect is unavoidable when taking photos through glasses or other transparent materials, which introduces undesired information into pictures. Hence, removing the influence of reflection becomes a key problem in computer vision. One of the main obstacles of recent learning based approaches is the lacking of realistic training data. To address this issue, we introduce a new dataset synthesis method as well as a novel neural network architecture for single image reflection removal. First, we make use of the polarization characteristics of light into the synthesis of datasets, so as to obtain more realistic and diversified training dataset POL. Then, we design a novel Progressive Polarization based Reflection Removal Network (P 2 R 2 Net), which preliminary estimates the coarse background layer to guide the final reflection removal. We demonstrate that our method performs better than the state-of-the-art single image reflection removal methods through quantitative and qualitative experimental comparisons. Specifically, the average PSNR of our restored images selected from three representative benchmark datesets: "Real20", " SI R 2 " and "Nature" is improved at least 0.49 compared with existing methods and reaches to 24.52. [ABSTRACT FROM AUTHOR]