“Do I Know You?” Altering hairstyle affects facial recognition

Bartel, Sara; Toews, Kelsi; Gronhovd, Lennae; Prime, Steven L.

doi:10.1080/13506285.2017.1394412

Cited by 2 publications

(5 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hair is a saliency feature of the head that may change in different situations. Even if there is sufficient information in the internal features of the face for recognition, manipulating the hair can harm the performance [172,173] . Moreover, facial synthesis and retrieval systems often use hair as an important cue [174,175] to improve the quality of generated images.…”

Section: Facial Feature Annotationmentioning

confidence: 99%

Facial-sketch Synthesis: A New Challenge

et al. 2022

View full text Add to dashboard Cite

This paper aims to conduct a comprehensive study on facial-sketch synthesis (FSS). However, due to the high cost of obtaining hand-drawn sketch datasets, there is a lack of a complete benchmark for assessing the development of FSS algorithms over the last decade. We first introduce a high-quality dataset for FSS, named FS2K, which consists of 2 104 image-sketch pairs spanning three types of sketch styles, image backgrounds, lighting conditions, skin colors, and facial attributes. FS2K differs from previous FSS datasets in difficulty, diversity, and scalability and should thus facilitate the progress of FSS research. Second, we present the largest-scale FSS investigation by reviewing 89 classic methods, including 25 handcrafted feature-based facial-sketch synthesis approaches, 29 general translation methods, and 35 image-to-sketch approaches. In addition, we elaborate comprehensive experiments on the existing 19 cutting-edge models. Third, we present a simple baseline for FSS, named FSGAN. With only two straightforward components, i.e., facial-aware masking and style-vector expansion, our FSGAN surpasses the performance of all previous state-of-the-art models on the proposed FS2K dataset by a large margin. Finally, we conclude with lessons learned over the past years and point out several unsolved challenges. Our code is available at https://github.com/DengPingFan/FSGAN.

show abstract

Section: Facial Feature Annotationmentioning

confidence: 99%

Facial-sketch Synthesis: A New Challenge

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Interestingly, hair usually draws attention when initially encoding a face (Bartel et al, 2018), and therefore, when hair becomes especially diagnostic (as in Experiment 1), the processing is moved away from the central aspects of the face and towards the hair instead to the detriment of the other features. This is in line with what the weighted representation model would assume; as we try to be parsimonious with our encoding of a face, we will initially encode what is seen as most diagnostic about a face (in this case hair seems especially important) with the assumption that encoding that distinctive feature will be to the detriment of other features (Devue, in prep).…”

Section: Discussionmentioning

confidence: 93%

“…As the weighted representation develops over time, it may be that hair is seen as diagnostic for both typical and distinctive faces because it is initially helpful for recognition when first encoding and learning a face. Some evidence suggests hair can be important for recognizing newly learned faces and unfamiliar faces (Bartel et al, 2018;Toseeb et al, 2012). Although others debate, that hair is not essential for our representation of faces (Murphy et al, 2014).…”

Section: Discussionmentioning

confidence: 99%

“…However, when participants were familiar with the same two identities, they successfully sorted the images into two sets of identities (Jenkins et al, 2011). Additionally, when initially learning a face, if hair is changed or removed in between learning and test phases, it significantly impairs recognition performance for these faces (Bartel et al, 2018;Toseeb et al, 2012).…”

Section: Qualitatively Different Processes? Familiar and Unfamiliar Face Processingmentioning

confidence: 99%

“…While the averaging hypothesis can explain how changes in viewpoint, lighting, expression, and tilt information can be ignored when familiarisation occurs; it does not account for the role of external features on familiar face recognition. It may be that during initial learning, external THE EFFECTS OF CONTEXT ON DIAGNOSTICITY OF FACIAL 13 FEATURES information along with internal information is encoded based more on what features are seen as most diagnostic for that face rather than averaging out all information that is variable when viewing a face (Bartel et al, 2018;Ellis et al, 1979).…”

Section: Current Theoriesmentioning

confidence: 99%

See 2 more Smart Citations

Saliency in Context: The Effect of Context on the Diagnosticity of Facial Features

Reedy¹

View full text Add to dashboard Cite

<p>How might faces we have learned be represented in our memory? Researchers believe that our memory for faces is based on building a robust averaged representation comprised of the stable aspects of the face (i.e., eyes, nose, mouth). However, anecdotal evidence suggests this one size fits all approach to face representations may not be correct. A new theory suggests our representation for faces is instead based on a dynamic weighting, wherein what is seen as most diagnostic during learning will be encoded to a greater extent than other features in the face. One factor that may be especially important for a weighted representation is the context in which a face is initially viewed. Dependent on the context of learning, certain features may appear more distinctive than others and therefore be deemed diagnostic and receive representational weight. The current study had participants learn four faces with one manipulated to appear distinctive in the experimental context by having a unique hair colour (Experiment 1), or eye colour (Experiment 2) compared to the other faces. Participants then completed a recognition task where the feature of interest (i.e., hair or eye colour) was either available or unavailable (i.e., bald and eye closed conditions) for recognition. Findings suggested recognition was disrupted when the diagnostic feature was unavailable compared to when that feature was available, across both distinctive and typical faces. Interestingly, Experiment 2 showed a distinctiveness performance advantage compared to Experiment 1, most likely because neighbouring features may be more diagnostic than others during recognition. In addition, further exploratory analysis showed the order of the test could further affect what was encoded.</p>

show abstract

“Do I Know You?” Altering hairstyle affects facial recognition

Cited by 2 publications

References 17 publications

Facial-sketch Synthesis: A New Challenge

Facial-sketch Synthesis: A New Challenge

Saliency in Context: The Effect of Context on the Diagnosticity of Facial Features

Contact Info

Product

Resources

About