Adaptive Irregular Graph Construction-Based Salient Object Detection

Zhou, Yuan; Zhang, Tianhao; Huo, Shuwei; Hou, Chunping; Kung, Sun‐Yuan

doi:10.1109/tcsvt.2019.2904463

Cited by 13 publications

(5 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [26], a fourlayer graph is constructed using multi-scale segmentation, and a third rank is executed with obtained foreground probability as features. [27] substitutes the conventional k-regular graph with an adaptive irregular graph and proposes a new seeding strategy. [28] considers regionally spatial consistency, and connects all potential foreground nodes and background nodes respectively.…”

Section: Related Workmentioning

confidence: 99%

Saliency Detection via Manifold Ranking on Multi-Layer Graph

Wang,

Ning,

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Saliency detection is increasingly a crucial task in the computer vision area. In previous graph-based saliency detection, superpixels are usually regarded as the primary processing units to enhance computational efficiency. Nevertheless, most methods do not take into account the potential impact of errors in superpixel segmentation, which may result in incorrect saliency values. To address this issue, we propose a novel approach that leverages the diversity of superpixel algorithms and constructs a multi-layer graph. Specifically, we segment the input image into multiple sets by different superpixel algorithms. Through connections within and connections between these superpixel sets, we can mitigate the errors caused by individual algorithms through collaborative solutions. In addition to spatial proximity, we also consider feature similarity in the process of graph construction. Connecting superpixels that are similar in feature space can force them to obtain consistent saliency values, thus addressing challenges brought by the scattered spatial distribution and the uneven internal appearance of salient objects. Additionally, we use the two-stage manifold ranking to compute the saliency value of each superpixel, which includes a background-based ranking and a foreground-based ranking. Finally, we employ a mean-field-based propagation method to refine the saliency map iteratively and achieve smoother results. To evaluate the performance of our approach, we compare our work with multiple advanced methods in four datasets quantitatively and qualitatively.

show abstract

Section: Related Workmentioning

confidence: 99%

Saliency Detection via Manifold Ranking on Multi-Layer Graph

Wang,

Ning,

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Initially, we detect the salient objects in a color image using conventional methods [36]- [38] or CNN-based [39]- [44] methods. However, some objects are very difficult to detect, irrespective of the method used.…”

Section: B Two-modality Salient Object Detectionmentioning

confidence: 99%

HRTransNet: HRFormer-Driven Two-Modality Salient Object Detection

Tang

Liu

Tan

et al. 2023

IEEE Trans. Circuits Syst. Video Technol.

View full text Add to dashboard Cite

High-Resolution Transformer (HRFormer) can maintain high-resolution representation and share global receptive fields. It is friendly towards salient object detection (SOD) in which the input and output have the same resolution. However, two critical problems need to be solved for two-modality SOD. One problem is two-modality fusion. The other problem is the HRFormer output's fusion. To address the first problem, a supplementary modality is injected into the primary modality by using global optimization and an attention mechanism to select and purify the modality at the input level. To solve the second problem, a dual-direction short connection fusion module is used to optimize the output features of HRFormer, thereby enhancing the detailed representation of objects at the output level. The proposed model, named HRTransNet, first introduces an auxiliary stream for feature extraction of supplementary modality. Then, features are injected into the primary modality at the beginning of each multi-resolution branch. Next, HRFormer is applied to achieve forwarding propagation. Finally, all the output features with different resolutions are aggregated by intrafeature and inter-feature interactive transformers. Application of the proposed model results in impressive improvement for driving two-modality SOD tasks, e.g., RGB-D, RGB-T, and light field SOD.https://github.com/liuzywen/HRTransNet

show abstract

“…In the content-base category, a lesion region in a radiograph, CT, or MRI gray image is commonly described by a rectangle ROI, which center and diameter (or diagonal) were estimated by unsupervised image processing methods, e.g., histogram thresholding, image transformation, pixel clustering or template-matching [15,23,19]. The regionbased category aimed to improve region-matching performance to the desired interest object, in which some region saliency strategies [36], e.g., attention window or supervision information [12,33,28], were incorporated into ROI selection by manual or semi-manual [21,2]. From then on, especially at the increasing requirements of interest object annotation, retrieval, detection and positioning, ROI selection and pixel extraction were extended to a wider scope of applications, e.g., tricolor natural images or multi-spectral RS images, even together with complex backgrounds.…”

Section: Introductionmentioning

confidence: 99%

Non-rigid Interest Object Extraction Using Minimum Convex Coverage

Yang,

Zhang,

Zhang

et al. 2023

Preprint

View full text Add to dashboard Cite

Interest object (IO) extraction, or called region of interest (ROI) selection, is a fundamental task in remote sensing (RS), computer vision, and machine learning. Due to the non-rigidity, e.g., shapelessness and color-uncertainty, of interest objects like fires, smokes, clouds, or top-view tree canopies in a RS image, an unideal results usually occurs in IO detection tasks because the model-training is carried on the pixels drawn from an inaccurate ROI, e.g., a bounding-box or convex region. Although it is simple, it fails to capture a complete interest object or sub-object, especially at the requirement of high-precision. In this work, we focus on the non-convex ROI and develop a method for the accurate pixel extraction. To speed pixel extraction, the given non-convex ROI is first decomposed into a minimum convex coverage, and then the ROI pixels can be quickly extracted out from the coverage by using convex operation. For the decomposition , we provide two strategies, named inner decomposition and outer decomposition. Computationally, batch-decision for pixel selection and logical operation for the decomposition are also presented to speed extraction. Theoretically , the existence of minimum coverage for both decompositions is proved. Finally, extensive experimental verifications are carried out on public and our collected RS images, and the results show the superiority of our method in terms of ease of use, rapidity and pixel-precision.

show abstract

Adaptive Irregular Graph Construction-Based Salient Object Detection

Cited by 13 publications

References 47 publications

Saliency Detection via Manifold Ranking on Multi-Layer Graph

Saliency Detection via Manifold Ranking on Multi-Layer Graph

HRTransNet: HRFormer-Driven Two-Modality Salient Object Detection

Non-rigid Interest Object Extraction Using Minimum Convex Coverage

Contact Info

Product

Resources

About