Accuracy and speed of material categorization in real-world images

Abstract

It is easy to visually distinguish a ceramic knife from one made of steel, a leather jacket from one made of denim, and a plush toy from one made of plastic. Most studies of material appearance have focused on the estimation of specific material properties such as albedo or surface gloss, and as a consequence, almost nothing is known about how we recognize material categories like leather or plastic. We have studied judgments of high-level material categories with a diverse set of real-world photographs, and we have shown (Sharan, 2009) that observers can categorize materials reliably and quickly. Performance on our tasks cannot be explained by simple differences in color, surface shape, or texture. Nor can the results be explained by observers merely performing shape-based object recognition. Rather, we argue that fast and accurate material categorization is a distinct, basic ability of the visual system.

Action observation: the less-explored part of higher-order vision

“…reported 30 ms to reach 75% correct in distinguishing an animal from a non-animal, Kovacs49 et al . 30 ms to reach 75% correct in shape discrimination, Green and Oliva51 50 ms to reach 75% threshold in basic scene categorization and Sharan et al 52. 40 ms to reach 80% correct in material categorization.…”

Processes of believing: Where do they come from? What are they good for?

“…Because of this high speed, most sensory information is processed non-consciously in the brain; only some bits of information enter consciousness. Accurate material categorization of real-world images occurs as fast as 30 ms but increases in accuracy with longer exposure times, up to 120 ms ( Sharan et al , 2014; van Gaal et al , 2011). As soon as information reaches the primary visual cortex, early automatic processing in the 100ms range affects information transmission further downstream ( Nortmann et al , 2015).…”

Artificial bee colony algorithm for content‐based image retrieval

“…In order to test the effectiveness of the proposed algorithm, the traditional k‐means, k‐nearest neighbors (KNN) methods and other approaches based on biologically inspired algorithms designed for CBIR, including the GA, PSO, BA, and artificial algae algorithm (AAA), were used for comparison with the CBIR results that were obtained from a number of different images. The experiment was conducted on the standard image dataset obtained from the Flickr Material Database . The five material surface categories, including leather, metal, stone, water, and wood, consisting of 10 images in each category were selected and used in this experiment, as shown in Figure .…”

“…The experiment was conducted on the standard image dataset obtained from the Flickr Material Database. 45 The five material surface categories, including leather, metal, stone, water, and wood, consisting of 10 images in each category were selected and used in this experiment, as shown in Figure 4. All methods in this experiment were programmed in C++, and all experiments were run on a PC with an Intel Core i7 CPU, 2.8 GHz, and 16 GB memory.…”