Salient object detection of surface defects is one of the surface defect detection tasks, which aims at highlighting the defect regions from the surface of strip steel, magnetic tale, road, and so on. However, the performance of existing methods degrades dramatically when dealing with complex scenarios, such as low contrast of defect regions and various defect shapes. Therefore, in this article, we propose a novel saliency model, namely, localizing, focus, and refinement network (LFRNet), which consists of the semantic-guided localizing module, the context-driven focus module, and the edge-aware refinement (ER) module. First, the semantic-guided localizing module deploys the graph reasoning (GR) unit and the global attention (GA) unit to localize the potential defect regions from a global view. Second, the context-driven focus module employs the split context (SC) unit and the mutual attention (MA) unit to perform the identification process via the introduction of spatial detail features. Finally, to further improve the accuracy of the detection results, we deploy the ER module, which introduces the boundary cues via the edge generation (EG) unit and aggregates the localizing result, the focus results, and the edge information into the high-quality detection map. Extensive experiments on four public defect datasets clearly show the effectiveness and superiority of the proposed LFRNet, where the LFRNet obtains an improvement of 4.1%, 5.7%, 1.0%, and 0.8% on F-measure (FM), weighted FM (WF), E-measure (EM), and structure-measure (SM), respectively, compared with the top-level method: AEP.

In the domain of cross-view geo-localization, the challenge lies in accurately matching images captured from distinct perspectives, such as aerial drone imagery and satellite imagery of the same geographical location. Existing methods predominantly concentrate on minimizing distances between feature embeddings in the representational space, inadvertently overlooking the significance of reducing embedding redundancy. This oversight potentially hampers the extraction of diverse and distinctive visual patterns critical for precise localization. This work argues that minimizing embedding redundancy is a pivotal factor in enhancing a model’s ability to discriminate diverse scene characteristics. To support this claim, we introduce a straightforward yet effective regularization technique, termed Dynamic Weighted Decorrelation Regularization (DWDR). DWDR serves to actively promote the learning of orthogonal feature channels within neural networks. By dynamically adjusting weights, DWDR targets the minimization of inter-channel correlations, guiding the correlation matrix towards diagonality, indicative of independence among channels. The dynamic weighting mechanism adaptively prioritizes the decorrelation of channels that remain highly correlated throughout training. Additionally, we devise a symmetrical sampling strategy for cross-view scenarios to ensure that the training examples are balanced across different imaging platforms in a batch. Despite its simplicity, the integration of DWDR and the proposed sampling scheme yields remarkable performance across four extensive benchmark datasets: University-1652, CVUSA, CVACT, and VIGOR. Notably, in stringent conditions, such as when constrained to exceedingly compact feature dimensions of 64, our methodology significantly outperforms conventional baselines, thereby affirming its efficacy and robustness under challenging constraints.

The reconstruction of the floorplan for a building requires the creation of a two-dimensional floorplan from a 3D model. This task is widely employed in interior design and decoration. In reality, the structures of indoor environments are complex with much clutter and occlusions, making it difficult to reconstruct a complete and accurate floorplan. It is well known that a suitable dataset is a key point to drive an effective algorithm, while existing datasets of floorplan reconstruction are synthetic and small. Without reliable accumulations of real datasets, the robustness of methods to real scene reconstruction is weakened. In this paper, we first annotate a large-scale realistic benchmark, which contains RGBD image sequences and 3D models of 80 indoor scenes with more than 10,000 square meters. We also introduce a framework for the floorplan reconstruction with mesh-based point cloud normalization. The loose-Manhattan constraint is performed in our optimization process, and the optimal floorplan is reconstructed via constraint integer programming. The experimental results on public and our own datasets demonstrate that the proposed method outperforms FloorNet and Floor-SP.

Video compression artifacts arise from quantization applied in the frequency domain. Video quality enhancement aims to reduce such compression artifacts and reconstruct a visually pleasant result. While existing methods effectively reduce artifacts in the spatial domain, they often overlook the rich frequency domain information, especially in addressing multi-scale compression artifacts. This work introduces a frequency-domain upsampling strategy within a multi-scale framework, specifically designed to focus on high-frequency details rather than simply blending neighboring pixels during the upsampling process. Our proposed hierarchical frequency-based upsampling and refinement neural network (HFUR) consists of two modules: implicit frequency upsampling (ImpFreqUp) and hierarchical and iterative refinement (HIR). ImpFreqUp exploits the DCT-domain prior derived through an implicit DCT transform, and accurately reconstructs the DCT-domain signal via a coarse-to-fine transfer. Additionally, HIR is introduced to facilitate cross-collaboration and information compensation between the scales, further refining the feature maps and promoting the visual quality of the final output.We demonstrate the effectiveness of the proposed modules via ablation experiments and visualized results. Experimental results demonstrate that HFUR outperforms the state-of-the-art methods up to 0.13dB/0.17dB on both constant bit rate and constant QP modes. The code is available at https://github.com/zqqqyu/HFUR.

Real-time dense SLAM techniques aim to reconstruct the dense three-dimensional geometry of a scene in real time with an RGB or RGB-D sensor. An indoor scene is an important type of working environment for these techniques. The planar prior can be used in this scenario to improve the reconstruction quality, especially for large low-texture regions that commonly occur in an indoor scene. This article fully explores the planar prior in a dense SLAM pipeline. First, we propose a novel plane detection and segmentation method that runs at 200 Hz on a modern graphics processing unit. Our algorithm for constructing global plane constraints is very efficient; hence, we use it in the process of each input frame for the camera pose estimation while maintaining the real-time performance. Second, we propose herein a plane-based map representation that greatly reduces the memory footprint of plane regions while keeping the geometric details on planes. The experiments reveal that our system yields superior reconstruction results with planar information running at more than 30 fps. Aside from speed and storage improvements, our technique also handles the low-texture problem in plane regions.