IEEE Transactions on Pattern Analysis and Machine Intelligence 24(6):748-763, 2002 Recognizing Imprecisely Localized, Partially Occluded and Expression Variant Faces from a Single Sample per Class

Citations in the corpus

• Face recognition in hyperspectral images (2003, methodology): “A method [12] that divides the face into regions for isolated analysis can tolerate up to 1/6 face occlusion without losing accuracy.”
• Face recognition: component-based versus global approaches (2003, methodology): “A probabilistic approach using part-based matching has been proposed in [18] for expression invariant and occlusion tolerant recognition of frontal faces.”
• Compound facial expressions of emotion (2014, result): “We also provide comparative results against a local-based approach as in (38).”
• Domain Adaptation with Multiple Sources (2008, background): “This problem arises in a variety of applications in natural language processing [4] [7] [10], speech processing [8, 9, 11, 13–15], computer vision [12], and many other areas.”
• Maximum Correntropy Criterion for Robust Face Recognition (2011, methodology): “To extract robust facial features, a number of algorithms, such as modular Eigenspaces [4], modular linear regression based classification (LRC) [5], eigenimages [6], [7], [8], minimax probability machine [9], reconstructive and discriminative subspace method [10], and associative memories [11], [12”
• Semantic Photo Synthesis (2006, background): “Literal patch matching is necessary because the Arc de Triomphe and many other objects are not yet recognized with semantic labels. images from a single face by applying stochastic affine warp perturbations to the face (see [Mar02] and [SP98]).”
• Occlusion Aware Facial Expression Recognition Using CNN With Attention Mechanism (2019, background): “To determine the patches on the face, existing works either divide the facial image into several uniform parts ([11]–[13]), or get the patches around the facial landmarks ( [14], [15]), or get the patches by a sampling strategy [16], or explicitly detect the occluders [13], [17], [18].”
• Past, Present, and Future of Face Recognition: A Review (2020, background): “Among these factors, we can cite the database size, low or high lighting, presence of noise or blur, disguises, partial occlusion, and certain secondary factors that are often common, unavoidable, and very challenging [18].”
• Partial Face Recognition: Alignment-Free Approach (2013, methodology): “A simple way is to divide the aligned face image into several sub-regions [34], [35], [36], [37], match each sub-region and then fuse the matching results.”
• Linear discriminant analysis for the small sample size problem: an overview (2014, unknown): “AR [32] Contains over 4,000 color images of 126 people’s faces (70 men and 56 women).”
• Dynamics of facial expression: recognition of facial actions and their temporal segments from face profile image sequences (2006, background): “Martinez [33] has shown that, by taking into account facial asymmetry caused by certain emotion, expression-invariant face recognition can be achieved.”
• Learning a similarity metric discriminatively, with application to face verification (2005, background): “Others have advocated warping-based normalization algorithms to maximally reduce the variations of appearance due to pose [10].”
• Multiple Nose Region Matching for 3D Face Recognition under Varying Facial Expression (2006, background): “Martinez has also worked on 2D face recognition in the context of facial expressions and noted that " different facial expressions influence different parts of the face more than others " [22].”
• Domain adaptation and sample bias correction theory and algorithm for regression (2014, background): “A more challenging problem of domain adaptation arises in a variety of applications, including natural language processing, speech processing, or computer vision [12, 4, 17, 18, 29, 30, 21].”
• A Comparative Study of Local Matching Approach for Face Recognition (2007, background): “The final classification was based on the linear combination of the probabilities of the six local regions [29].”