On the Significance of Real-World Conditions for Material Classification

By Eric Hayman, Barbara Caputo, Mario Fritz, Jan-Olof Eklundh. 2004

401 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Industrial and Manufacturing Engineering

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 Industrial and Manufacturing Engineering

Citations in the corpus

Describing Textures in the Wild (2013, methodology): “Surpris-ingly, these descriptors outperform specialized state-of-the-art texture representations not only in recognizing our de-scribable attributes, but also in a variety of datasets for material recognition, achieving an accuracy of 63.3% on FMD and 67.5% on KTH-TIPS2-b dataset (Sect.”
Material recognition in the wild with the Materials in Context Database (2014, background): “Later, databases with more diverse examples from each material category began to appear, such as KTH-TIPS [10] and KTH-TIPS2 [2], and led explorations of how to generalize from one example of a material to another—from one sample of wood to a completely different sample, for instance.”
Deep filter banks for texture recognition and segmentation (2015, background): “The third texture dataset is KTH-TIPS2b [5, 17], containing a number of example images for each of four samples of 11 material categories.”
Classifying materials in the real world (2010, background): “1 Our results were state-of-the-art when first presented in [1], but Broadhurst [6”
The Pyramid Match Kernel: Efficient Learning with Sets of Features (2007, methodology): “This type of representation was explored for texture recognition using nearest-neighbors (Leung and Malik, 2001; Hayman et al., 2004), and more recently has been shown for classification of object categories using SVMs, Na¨ıve Bayes classifiers, and a probabilistic Latent Semantic Analysis…”
A Statistical Approach to Texture Classification from Single Images (2005, background): “For instance, Support Vector Machines (Cristianini and Shawe-Taylor, 2000; Hayman et al., 2004; Kim et al., 2002; Schólkopf and Smola, 2002), and perhaps more appropriately Relevance Vector Machines (Tipping, 2001), are both capable of reducing the number of models while providing good generalisatio”
A Statistical Approach to Texture Classification from Single Images (2004, background): “For instance, Support Vector Machines (Cristianini and Shawe-Taylor, 2000; Hayman et al., 2004; Kim et al., 2002; Schólkopf and Smola, 2002), and perhaps more appropriately Relevance Vector Machines (Tipping, 2001), are both capable of reducing the number of models while providing good generalisatio”
Scattering Representations for Recognition (2013, background): “53% for a dictionary of textons [HCFE04], 1.”
Deep Filter Banks for Texture Recognition, Description, and Segmentation (2015, background): “This choice trades-off the systematic sampling of illumination and viewpoint variations existing in datasets such as CUReT, KTH-TIPS, Outex, and Drexel to capture real-world variations, reducing the gap with applications.”
Local Higher-Order Statistics (LHS) for Texture Categorization and Facial Analysis (2012, methodology): “Most of the other recent methods, driven by the success of earlier texton based texture classification method [2] and recent advances in the field of object category classification, adopt bag-of-words models to represent textures as distributions of local textons [7, 17–24].”
Invariant Scattering Convolution Networks (2013, background): “53 percent for a dictionary of textons [15], 1.”
A Statistical Approach to Material Classification Using Image Patch Exemplars (2009, background): “In turn, [4], [7], [21] In the learning stage, training images are convolved with a chosen ﬁlter bank to generate ﬁlter responses.”
Pairwise Rotation Invariant Co-Occurrence Local Binary Pattern (2014, methodology): “The KTH-TIPS data set [35] consists of 10 texture classes.”
Class-specific material categorisation (2005, methodology): “While the choice between four state-of-the-art descriptors [24, 25, 20, 16] has little effect, using Support Vector Machines (SVMs) over Nearest Neighbour classifiers proves beneficial, as it did for material identification in [12].”