Identification of spatially quantised tachistoscopic images of faces: How many pixels does it take to carry identity?

By Talis Bachmann. 1991

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

Citations in the corpus

A Study and Comparison of Human and Deep Learning Recognition Performance under Visual Distortions (2017, background): “Similarly, Bachmann [11] show that human subjects can recognize low resolution faces.”
Analyzing Semantic Segmentation Using Hybrid Human-Machine CRFs (2013, background): “[2] show that humans can reliably recognize faces in 16 × 16 images, and Oliva et al.”
80 Million Tiny Images: A Large Data Set for Nonparametric Object and Scene Recognition (2008, background): “Scene recognition Studies on face perception [1], [21] have shown that only 16× 16 pixels are needed for robust face recognition.”
Face recognition: A literature survey (2003, background): “However, it has been demonstrated that the image size can be very small for holistic face recognition: 12 × 11 for the subspace LDA system [Zhao et al. 1999], 14×10 for the PDBNN system [Lin et al. 1997], and 18 × 24 for human perception [Bachmann 1991].”
The involvement of distinct visual channels in rapid attention towards fearful facial expressions (2005, background): “Low spatial frequency (LSF) information, on the other hand, provides coarse scale cues for global shadow-related changes in the face, which are informative about configural structure, such as the relative positions of the eyes, nose, and mouth (e.g., Bachmann, 1991; Costen, Parker, & Craw, 1996).”
Robust representations for face recognition: the power of averages. (2005, background): “…is, in fact, consistent with research on spatial scale in face recognition, which suggests that identity tends to be carried at low spatial scales (Bachmann, 1991; Harmon & Julesz, 1973; though see Schyns & Oliva, 1997, for an argument that extraction of information from different spatial scales…”
Human and DNN Classification Performance on Images With Quality Distortions (2019, background): “Torralba et al. (2008) show that human subjects can accurately recognize low-resolution images, and Bachmann (1991) show a similar result on blurred images of faces.”
Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Exploring Tiny Images: the Roles of Appearance and Contextual Information for Machine and Human Object Recognition (unknown): “Human accuracies have been studied in low resolution images for face recognition [36], [37], scene recognition [38], [39], [19], [40] and more recently for object detection [41], [19] and segmentations [19].”