Submitted to Ieee Transactions on Pattern Analysis and Machine Intelligence Learning to Detect Natural Image Boundaries Using Local Brightness, Color, and Texture Cues

By David R. Martin, Charless C. Fowlkes, Jitendra Malik

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Panoptic Segmentation (2018, methodology): “Written this way, RQ is the familiar F1 score [45] widely used for quality estimation in detection settings [33].”
Finely-grained annotated datasets for image-based plant phenotyping (2016, positive): “…when images are partially out-of-focus and of lower resolution (due to a larger ﬁeld of view) as is the case of the Arabidopsis data, learning-based methods to boundary estimation (i.e. a learned edge detector as in Martin et al. (2004)) have been shown to perform better (Pape and Klukas, 2015).”
Holistically-Nested Edge Detection (2015, methodology): “After decades of research, there have emerged a number of properties that are key and that are likely to play a role in a successful system: (1) carefully designed and/or learned features (Martin et al. 2004; Dollár et al. 2006), (2)”
Fast Edge Detection Using Structured Forests (2014, background): “F”
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation (2015, methodology): “Hence they propose to complement the mIoU metric with a boundary measure based on the Berkeley contour matching score commonly used to evaluate unsupervised image segmentation quality [58].”
On the Binding Problem in Artificial Neural Networks (2020, background): “, 2017), to learn boundary detection (Martin et al., 2004; Hoiem et al., 2011), or as a means of top-down feedback (Mobahi et al.”
Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference (2020, methodology): “optimal dataset F-measure (odsF) [40] for edge detection, mean intersection over union (mIoU) for semantic segmentation, human parts and saliency, and finally mean error (mErr) for surface normals.”
Unsupervised Learning of Edges (2015, methodology): “Specifically, in datasets such as BSDS [2], human annotators use their knowledge of scene structure and object presence to mark semantically meaningful edges.1 Moreover, recent edge detectors use ImageNet pre-training [5, 44].”
Multiscale Combinatorial Grouping for Image Segmentation and Object Proposal Generation (2015, methodology): “We found the results with all learning methods surprisingly similar, in agreement with the observation reported by [33].”
Pushing the Boundaries of Boundary Detection using Deep Learning (2015, background): “In particular, since human annotations come with some positional uncertainty, the standard evaluation protocol of Martin et al. (2004) allows for some slack in the predicted position of a pixel (a ﬁxed fraction of the image diagonal).”
Crisp Boundary Detection Using Pointwise Mutual Information (2014, background): “[15] define linear operators for a number of cues such as intensity, color and texture.”
Semantic Segmentation using Adversarial Networks (2016, methodology): “This measure extends the Berkeley contour matching score [17], a commonly used metric in segmentation, to semantic segmentation.”
Oriented edge forests for boundary detection (2014, background): “insufﬁcient for handling many natural scenes where local gradients are dominated by ﬁne scale clutter and texture arising from surface roughness and varying albedo. Modern boundary detectors, such as [19], have emphasized the importance of suppressing such responses by explicit oriented analysis of ”
Advanced Structured Prediction (2014, background): “F -measure is one of the most common performance measures for boundary detection (Martin et al., 2004).”
Attentive Single-Tasking of Multiple Tasks (2019, methodology): “For edge detection we use the BSDS500 [38] benchmark and its optimal dataset F-measure ( odsF ) [37].”