The advent of scribble-supervised learning has opened new frontiers in weakly supervised RGB-D salient object detection (SOD), representing a novel paradigm that significantly reduces annotation costs while maintaining detection efficacy. The SOD methods predominantly based on convolutional neural networks (CNNs) exhibit notable limitations in capturing global contextual dependencies and effectively utilizing hierarchical multiscale representations. Furthermore, conventional edge extraction approaches based on Canny edge detection demonstrate inherent limitations in preserving structural coherence across diverse object boundaries. To address these critical challenges, we propose an NESS-Net model, an innovative weakly supervised architecture that synergistically integrates our previously developed NAMLab framework with Swin-Transformer-based feature learning. The core contributions of our work are threefold: Firstly, building upon our prior work in perceptual-aware image segmentation, we employ the NAMLab algorithm—a hierarchical segmentation framework inspired by Nonsymmetry and Anti-packing pattern representation Model in the Lab color space (NAMLab)—to generate edge maps that better align with human visual perception. Secondly, we devise a U-shaped Edge Construction Module (ECM) that systematically refines edge features through progressive refinement of the NAMLab-generated edge priors. Thirdly, leveraging Swin-Transformer’s hierarchical attention mechanism, we put forward a lightweight Cross-Attention Fusion Module (CAFM) that establishes long-range dependencies across modalities while maintaining computational efficiency. This architectural innovation naturally extends to our Cross-Modal multi-scale Attention-weighting Module (CMAM), which explicitly models inter-modal relationships through transformer-based attention weighting. The extensive experimental results on nine benchmark RGB-D SOD datasets demonstrate that our proposed NESS-Net model outperforms all the current state-of-the-art scribble-supervised models and even achieves performance competitive to leading fully supervised models. The source code and pre-trained models will be made publicly available at https://github.com/EricLie-c/NESS-Net.

Recently, remote sensing object detection (RSOD) has attracted increasing attention. Despite advancements in existing methods, challenges like high computational costs, large object scale variations, and difficulties in detecting small objects still remain. Many studies have attempted to address these issues by expanding the reception field but still struggle with multi-scale targets. To address these challenges, we propose SPCNet, an efficient multi-scale convolutional network that enhances feature maps by combining serial convolutions with a pyramid structure. First, we have designed a lightweight Serial Pyramid Convolutional (SPC) block that uses depthwise separable convolutions with varying dilation rates to capture multi-scale features and a pyramid structure to mitigate feature information loss caused by stacked convolutions. This mechanism is particularly effective at capturing intricate spatial relationships and at focusing on salient regions. Additionally, we have developed a novel backbone module, named SPC Stage, which integrates multiple SPC blocks along with a parallel large-kernel Feed-Forward Network (FFN) to aggregate multi-scale and wide-range contextual information. To confirm its effectiveness, we have evaluated our method on three widely used datasets named DOTA-v1.0, DOTA-v1.5, and HRSC2016. The extensive experiments on these datasets show that our SPCNet achieves the state-of-the-art performance with minimal parameters (72.48% mAP, 78.42% mAP, and 70.60% mAP, respectively). Our code will be made publicly available on https://github.com/FitzroyWangzj/SPCNet.

Abstract This paper proposes a new method for hierarchical image segmentation based on the nonsymetry and anti‐packing pattern representation model (NAM) and the hierarchical density‐based spatial clustering of application with noise (HDBSCAN). The proposed framework consists of two phases. In the first phase, a super‐pixel generation algorithm base on NAM is proposed. In the second phase, instead of defining an affinity matrix to merge similar regions using spatial clustering, the distance matrix defined by different region features is directly fitted into an HDBSCAN clustering module in order to merge similar regions efficiently. Similar adjacent regions can be merged into larger ones progressively and form a segmentation dendrogram for image segmentation with the clustering module. The experiments show that the proposed algorithm has a comparable or even better performance compared to the state‐of‐the‐art hierarchical image segmentation algorithms while having much less time and memory consumption.

Abstract Region extraction is usually used by many computer vision tasks as a pre‐processing step to extract image features. However, how to efficiently extract effective regions remains a challenging problem. In this paper, inspired by the non‐symmetry and anti‐packing pattern representation model (NAM) and the FatRegion algorithm, a fast NAM‐based region extraction algorithm which is called FNRegion is proposed. A NAM‐based homogeneous block generation algorithm is first presented to represent an image as a combination of multiple homogeneous blocks, each of which is a square region with visually indistinguishable intra‐region colour difference. Then, these homogeneous blocks are merged into larger regions according to their colour and shape information. To group these regions into larger ones in order to progressively build a region tree, a distance function is defined using variety of regional information to measure the distance between adjacent regions. Also, a multi‐feature region merging algorithm with linear complexity both in time and space is presented.The proposed algorithm has been evaluated on multiple public datasets in comparison with the state‐of‐the‐art region extraction algorithms. The experimental results show that in the case of almost the same or even less running time as other fast region extraction algorithms, the proposed algorithm is able to extract higher‐quality regions.

Image segmentation is the foundation of high-level image analysis and image understanding. How to effectively segment an image into regions that are "meaningful" to the human visual perception and ensure that the segmented regions are consistent at different resolutions is still a very challenging issue. Inspired by the idea of the Nonsymmetry and Anti-packing pattern representation Model in the Lab color space (NAMLab) and the "global-first" invariant perceptual theory, in this paper, we propose a novel framework for hierarchical image segmentation. Firstly, by defining the dissimilarity between two pixels in the Lab color space, we propose an NAMLab-based color image representation approach that is more in line with the human visual perception characteristics and can make the image pixels fast and effectively merge into the NAMLab blocks. Then, by defining the dissimilarity between two NAMLab-based regions and iteratively executing NAMLab-based merging algorithm of adjacent regions into larger ones to progressively generate a segmentation dendrogram, we propose a fast NAMLab-based algorithm for hierarchical image segmentation. Finally, the complexities of our proposed NAMLab-based algorithm for hierarchical image segmentation are analyzed in details. The experimental results presented in this paper show that our proposed algorithm when compared with the state-of-the-art algorithms not only can preserve more details of the object boundaries, but also it can better identify the foreground objects with similar color distributions. Also, our proposed algorithm can be executed much faster and takes up less memory and therefore it is a better algorithm for hierarchical image segmentation.

The fractal image compression (FIC) algorithm is difficult to be widely used in real‐time applications due to the huge consumption of its encoding time. Inspired by the fact that in the decoding process, for any original image, a fixed point is generated by the iterations of the fractal codes, the authors propose a specific update search FIC (SUSFIC) algorithm, which uses a scale number to control the update times of the fractal codes and to find acceptable matching domain blocks rather than the best ones in the selected domain blocks pool. To further reduce the computation time, in their proposed algorithm, the image blocks created by the equidistant sampling in the range of the original blocks are used to replace themselves when calculating the correlation coefficients as the distances between the adjacent domain blocks. The experimental results presented show that their proposed SUSFIC algorithm has a significant improvement in encoding time under the premise of setting an appropriate search update threshold and maintaining image quality when compared with the state‐of‐the‐art FIC algorithms. Therefore, it is a better FIC algorithm.