Towards Interactive Image Segmentation by Dynamic and Iterative Spanning Forest

Barcelos, Isabela B.; Belém, Felipe; Miranda, Paulo A. V.; Falcão, Alexandre X.; Patrocínio, Zenilton Kleber G. do; Guimarães, Silvio Jamil Ferzoli

doi:10.1007/978-3-030-76657-3_25

Cited by 8 publications

(7 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For future endeavors, we intent to study different curves for establishing the number of seeds at each iteration. Similarly to [30], we also desire to investigate the applicability of ODISF for interactive object segmentation based on userdrawn markers.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Towards a Simple and Efficient Object-based Superpixel Delineation Framework

Belém

Cousty

Guimarães

et al. 2021

2021 34th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)

View full text Add to dashboard Cite

Superpixel segmentation methods are widely used in computer vision applications due to their properties in border delineation. These methods do not usually take into account any prior object information. Although there are a few exceptions, such methods significantly rely on the quality of the object information provided and present high computational cost in most practical cases. Inspired by such approaches, we propose Object-based Dynamic and Iterative Spanning Forest (ODISF), a novel object-based superpixel segmentation framework to effectively exploit prior object information while being robust to the quality of that information. ODISF consists of three independent steps: (i) seed oversampling; (ii) dynamic path-based superpixel generation; and (iii) object-based seed removal. After (i), steps (ii) and (iii) are repeated until the desired number of superpixels is finally reached. Experimental results show that ODISF can surpass state-of-the-art methods according to several metrics, while being significantly faster than its object-based counterparts.

show abstract

Section: Discussionmentioning

confidence: 99%

“…As one may note, the contrast information is also insufficient since low contrast regions are present within the whole image. Thus, inspired by [30], we propose an object-based seed relevance criterion V 2 (Eq. 3)…”

Section: Object-based Seed Removalmentioning

confidence: 99%

Towards a Simple and Efficient Object-based Superpixel Delineation Framework

Belém

Cousty

Guimarães

et al. 2021

2021 34th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)

View full text Add to dashboard Cite

show abstract

“…The Interactive Dynamic and Iterative Spanning Forest (iDISF) is an interactive segmentation method derived from the DISF superpixel computation method [8]. The DISF method is able to compute very accurate superpixel delineation, but it does not relate the superpixels to any object of interest.…”

Section: A Interactive Dynamic and Iterative Spanning Forestmentioning

confidence: 99%

“…The final segmentation Seg I (S O , S e B ) is given by iteratively computing the optimumpath forest [10] rooted on the given set of seeds, filtering the least relevant seeds at each step. The seed removal criteria are detailed in [8]. A sample of a cell image segmentation produced with iDISF is illustrated in Figure 2.…”

Section: A Interactive Dynamic and Iterative Spanning Forestmentioning

confidence: 99%

Single-Shot Object Detection and Supervised Image Segmentation for Analysing Cell Images Obtained by Immunohistochemistry

Costa,

Rodrigues,

Fonseca

et al. 2023

Anais Estendidos Da XXXVI Conference on Graphics, Patterns and Images (SIBRAPI Estendido 2023)

View full text Add to dashboard Cite

Analyzing cell images and identifying them correctly is a fundamental task in the immunohistochemical exam. In this paper we propose a novel method to segment FoxP3+ Regulatory T cells (Treg) images automatically, in order to assist healthcare professionals in the task of identifying and counting potentially cancerous cells. The proposed method relies on combining an object detection network, which is tailor-made for microscopy images, with a marker-based image segmentation method to produce the final segmentation, while requiring only a 50x50 training patch to do so. Our pipeline consists on predicting the location of the cells, applying morphological operations on the prediction weights to transform them into markers, and finally using the segmentation method iDISF to generate high quality segmentations. We also propose a new FoxP3+ Treg cells dataset containing 10 high resolution images, with a qualitative and quantitative analysis of our segmentation methods for this dataset.

show abstract

“…The Image Foresting Transform (IFT) [11] is a framework whose effectiveness in object delineation has been reported in several works [13,6,3,9]. When a set of representative vertices (i.e., seeds) S ⊂ V is provided, the algorithm builds trees with optimum path-cost from their seed s ∈ S to any p ∈ V \ S through path concatenation.…”

Section: Image Foresting Transformmentioning

confidence: 99%