Scene text erasing, which replaces text regions with reasonable content in natural images, has drawn significant attention in the computer vision community in recent years. There are two potential subtasks in scene text erasing: text detection and image inpainting. Both subtasks require considerable data to achieve better performance; however, the lack of a large-scale real-world scene-text removal dataset does not allow existing methods to realize their potential. To compensate for the lack of pairwise real-world data, we made considerable use of synthetic text after additional enhancement and subsequently trained our model only on the dataset generated by the improved synthetic text engine. Our proposed network contains a stroke mask prediction module and background inpainting module that can extract the text stroke as a relatively small hole from the cropped text image to maintain more background content for better inpainting results. This model can partially erase text instances in a scene image with a bounding box or work with an existing scene-text detector for automatic scene text erasing. The experimental results from the qualitative and quantitative evaluation on the SCUT-Syn, ICDAR2013, and SCUT-EnsText datasets demonstrate that our method significantly outperforms existing state-of-the-art methods even when they are trained on real-world data.

Infrared image super-resolution (IRSR) is challenging due to weak structures and textures. While Mamba-based state-space models (SSMs) efficiently model long-range dependencies, their inherent block-wise processing disrupts spatial consistency, limiting direct IRSR applicability. We propose IRSRMamba, a novel Mamba-based framework overcoming this limitation via tailored structural and textural preservation. Integrated into a Mamba backbone, our key innovations are: 1) Wavelet Transform Feature Modulation (WTFM), enhancing multi-scale frequency-aware feature extraction to mitigate block-induced coherence loss; and 2) an SSMs-based Semantic Consistency Loss, enforcing cross-block alignment to restore fragmented context. IRSRMamba achieves superior global-local fusion, structural coherence, and fine-detail preservation. Experiments show state-of-the-art PSNR, SSIM, and perceptual quality on IR benchmarks, as well as robust generalization to remote sensing. This work establishes Mamba-based architectures as highly promising for high-fidelity IR image enhancement. Code is available at https://github.com/yongsongH/IRSRMamba.

We propose a GAN-based image compression method working at extremely low bitrates below 0.1bpp. Most existing learned image compression methods suffer from blur at extremely low bitrates. Although GAN can help to reconstruct sharp images, there are two drawbacks. First, GAN makes training unstable. Second, the reconstructions often contain unpleasing noise or artifacts. To address both of the drawbacks, our method adopts two-stage training and network interpolation. The two-stage training is effective to stabilize the training. Moreover, the network interpolation utilizes the models in both stages and reduces undesirable noise and artifacts, while maintaining important edges. Hence, we can control the trade-off between perceptual quality and fidelity without re-training models. The experimental results show that our model can reconstruct high quality images. Furthermore, our user study confirms that our reconstructions are preferable to state-of-the-art GAN-based image compression model. Our source code is available at https://github.com/iwa-shi/fidelity_controllable_compression

Feature extraction is essential for chemical property estimation of molecules using machine learning. Recently, graph neural networks have attracted attention for feature extraction from molecules. However, existing methods focus only on specific structural information, such as node relationship. In this paper, we propose a novel graph convolutional neural network that performs feature extraction with simultaneously considering multiple structures. Specifically, we propose feature extraction paths specialized in node, edge, and three-dimensional structures. Moreover, we propose an attention mechanism to aggregate the features extracted by the paths. The attention aggregation enables us to select useful features dynamically. The experimental results showed that the proposed method outperformed previous methods.

Image Super-Resolution (SR) is a fundamental task in computer vision and image processing. The challenge of super-resolving infrared (IR) or thermal images represents a key and persistent research frontier in the deep learning era. Different from visible-light SR, IRSR is challenged by the distinct properties of thermal images, which typically exhibit low contrast, limited high-frequency details, and sensor-specific noise. This survey provides a comprehensive review of Infrared Image Super-Resolution (IRSR), covering its applications, the inherent limitations of IR imaging hardware, and a detailed taxonomy of processing methodologies. In addition, we discuss the critical roles of datasets and evaluation metrics in the field. Finally, we highlight current technological gaps and identify promising future directions for the community. To keep pace with the rapid developments, we maintain an actively updated repository of relevant work at https://github.com/yongsongH/Infrared Image\_SR\_Survey.

We propose an end-to-end consistently trainable text detection method based on the Faster R-CNN. The original Faster R-CNN is an end-to-end CNN for fast and accurate object detection. By considering the characteristics of texts, a novel architecture that make use of its ability on object detection is proposed. Although the original Faster R-CNN generates region of interests (RoIs) by a region proposal network (RPN) using the feature map of the last convolutional layer, the proposed method generates RoIs by multiple RPNs using the feature maps of multiple convolutional layers. This method uses multiresolution feature maps to detect texts of various sizes simultaneously. To aggregate the RoIs, we introduce RoI-merge layer, and this layer enables to select valid RoIs from multiple RPNs effectively. In addition, a training strategy is proposed for realizing end-to-end training and making each RPN be specialized in text region size. Experimental results using ICDAR2013/2015 RRC test dataset show that the proposed Multi-RPN method improved detection scores and kept almost the same detection speed as compared to the original Faster R-CNN and recent methods.