Efficient, high-quality image contour detection

By Bryan Catanzaro, Bor-Yiing Su, Narayanan Sundaram, Yunsup Lee, Mark Murphy, Kurt Keutzer. 2009

143 citations recorded by the source corpus.

Image contour detection is fundamental to many image analysis applications, including image segmentation, object recognition and classification. However, highly accurate image contour detection algorithms are also very computationally intensive, which limits their applicability, even for offline batch processing. In this work, we examine efficient parallel algorithms for performing image contour detection, with particular attention paid to local image analysis as well as the generalized eigensolver used in Normalized Cuts. Combining these algorithms into a contour detector, along with careful implementation on highly parallel, commodity processors from Nvidia, our contour detector provides uncompromised contour accuracy, with an F-metric of 0.70 on the Berkeley Segmentation Dataset. Runtime is reduced from 4 minutes to 1.8 seconds. The efficiency gains we realize enable high-quality image contour detection on much larger images than previously practical, and the algorithms we propose are applicable to several image segmentation approaches. Efficient, scalable, yet highly accurate image contour detection will facilitate increased performance in many computer vision applications.

Subfields: Computer Vision and Pattern Recognition, Efficiency & Compression

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

In Efficiency & Compression

Citations in the corpus

Fast Edge Detection Using Structured Forests (2014, methodology): “Optionally, the independent edge predictions may then be combined using global reasoning [1], [41], [49], [2].”
The GPU Computing Era (2010, methodology): “Table 3 lists some representative applications along with the runtime speedups obtained for the whole application using CPUþGPU coprocessing over CPU alone, as measured by application developers.(12-22) The speedups using GeForce 8800, Tesla T8, GeForce GTX 280, Tesla T10, and GeForce GTX 285 range ”
Semantic Image Segmentation with Contextual Hierarchical Models (2015, background): “While some approaches such as [10], [11], [12], [13] can only be applied to edge detection problems and other approaches such as [14], [15], [16] are only designed for the image labeling problem, CHM can handle both problems equallywell without anymodification.”
Shape-based pedestrian parsing (2011, background): “This segmentation algorithm achieves state-of-the art performance on general purpose segmentation benchmarks and a fast implementation is available which exploits computation on the GPU to segment an image in a few seconds [6].”
Attribute Learning in Large-Scale Datasets (2010, methodology): “We represent each image using three types of normalized histogram features: (1) color histogram of of quantized RGB pixels using a codebook of size 50, (2) texture histogram of quantized SIFT descriptors at multiple levels using a codebook of size 1000 [10, 11], and (3) shape histogram of quantized ”
Structured Forests for Fast Edge Detection (2013, methodology): “Note that the GPU implementation [6] of [1] only achieves an ODS score of 0.”
Generalized Boundaries from Multiple Image Interpretations (2012, methodology): “Pb [1], [5] achieves an F-measure of 0.”
Boundary Detection Using F-Measure-, Filter- and Feature- (F3) Boost (2010, methodology): “The most powerful boundary detectors currently in use [8, 16, 17] rely on combining different cues for boundary detection, such as texture gradients, brightness/color gradients, or information extracted from spectral clustering.”