On seeing stuff: the perception of materials by humans and machines

By Edward H. Adelson. 2001

419 citations recorded by the source corpus.

Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, Geology

How researchers see this paper

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

In Computer Graphics and Computer-Aided Design

In Computer Vision and Pattern Recognition

In Geology

Citations in the corpus

Unsupervised Visual Representation Learning by Context Prediction (2015, background): “occur in a specific spatial configuration (if there is no specific configuration of the parts, then it is “stuff” [1]).”
Panoptic Segmentation (2018, background): “Questioning the wisdom of this trend, Adelson [1] elevated the importance of studying systems that recognize stuff – amorphous regions of similar texture or material such as grass, sky, road.”
COCO-Stuff: Thing and Stuff Classes in Context (2016, background): “(2) Size: Things occur at characteristic sizes with little variance, whereas stuff regions are highly variable in size [21, 2, 27].”
SUN database: Large-scale scene recognition from abbey to zoo (2010, background): “In this regard, scenes behave more like materials (or “stuff” [1]) in that they have unspecified spatial extent.”
Segment Anything (2023, background): “We did not impose semantic constraints for labeling objects, and annotators freely labeled both “stuff” and “things” [1].”
Context as Supervisory Signal: Discovering Objects with Predictable Context (2014, background): “an object with a well-defined shape, versus 2) that the image is best described as ‘stuff’ [26], i.”
Physics 101: Learning Physical Object Properties from Unlabeled Videos (2016, background): “There have been several early efforts to build computational vision systems with the physical knowledge of an early child [6], or specific physical properties, such as material, from the perspectives of human and machine perception [1, 17].”
Learning to Fuse Things and Stuff (2018, background): “On one hand, the best semantic segmentation networks [3, 43, 6] focus on dense tensor-to-tensor classification architectures that excel at recognizing stuff categories like roads, buildings, or sky [1].”
Contextual Models for Object Detection Using Boosted Random Fields (2004, background): “Ex tended stuff [1] may be a very important source of contextual information for other objec ts.”
Real-world illumination and the perception of surface reflectance properties. (2003, background): “This is consistent with the example of the pearlescent sphere above, and our previous report (Fleming, Dror, & Adelson, 2001).”
Recognition by association via learning per-exemplar distances (2008, background): “For example, Adelson divides the world into “things” (such as cars, people) and “stuff” (grass, pavement, ice cream, etc.) [1].”
Deep filter banks for texture recognition and segmentation (2015, background): “Hence, a significant amount of effort in the computer vision community has gone into recognizing texture via tasks such as texture perception [1, 2, 12, 13] and description [7, 11], material recognition [26, 36, 37], segmentation [20, 29], and even synthesis [9, 43].”
A parametric texture model based on deep convolutional features closely matches texture appearance for humans (2017, background): “Texture is an important material property whose perception is of adaptive value (Adelson, 2001; Fleming, 2014).”
Deep Filter Banks for Texture Recognition, Description, and Segmentation (2015, background): “r ubiquitousness and complementarity to other visual properties such as shape, have been 1 PASCAL VOC 2007 uses instead an interpolated version of mAP. studied in several contexts: texture perception [2, 3, 38, 40], description [36], material recognition [60, 60, 72, 87, 88, 97, 98], segmentation [2”
Automated annotation of coral reef survey images (2012, background): “As noted by Adelson [1], rather than trying to recognize things as found in most of the object recognition datasets, biological datasets often require identifying stuff.”
Optical Flow with Semantic Segmentation and Localized Layers (2016, background): “(3) Stuff [3] corresponds to classes that exhibit textural motion or objects like ‘buildings’ and ‘vegetation’ that may have a complicated 3D shape, exhibit complex parallax, and for which we have no compact motion representation.”