When is scene recognition just texture recognition
By L. Walker, Jitendra Malik. 2010
53 citations recorded by the source corpus.
Subjects were asked to discriminate scenes after very brief exposures (37-69 ms). Their performance was always above chance and increased with exposure duration, confirming that subjects can get the gist of a scene with one fixation. We propose that a simple texture analysis of the image can provide a useful cue towards rapid scene identification. Our model learns texture features across scene categories and then uses this knowledge to categorize new scenes. The texture analysis leads to similar categorizations and confusions as subjects with limited processing time. We conclude that a simple texture discrimination model mostly explains early scene identification.
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
- Luc Van Gool
- Antonio Torralba
- Ming-Hsuan Yang
- Dacheng Tao
- Ali Borji
- Xiaogang Wang
- Jitendra Malik
- Trevor Darrell
Citations in the corpus
- SUN Database: Exploring a Large Collection of Scene Categories (2014, background): “Whereas most computational work on scene and place recognition has used a limited number of semantic categories, representing typical indoor and outdoor settings [25, 16,32,50,29,4,46], access to large quantities of images on the Internet now makes it possible to build comprehensive datasets of imag”
- 80 Million Tiny Images: A Large Data Set for Nonparametric Object and Scene Recognition (2008, background): “distinct from studies investigating scene recognition usi ng very short presentation times [11], [30], [33], [ 34 ].”
- State-of-the-Art in Visual Attention Modeling (2013, methodology): “Renninger and Malik [94] applied Gabor filters to an input image and then extracted 100 universal textons selected from a training set using K-means clustering.”
- Human vs. Computer in Scene and Object Recognition (2014, methodology): “Texton histogram [11] is a 512D histogram by assigning each pixel’s set of filter responses to the nearest texton dictionary entry (from a universal texton dictionary [11])1.”