Face detection has been well studied for many years and one of remaining challenges is to detect small, blurred and partially occluded faces in uncontrolled environment. This paper proposes a novel contextassisted single shot face detector, named PyramidBox to handle the hard face detection problem. Observing the importance of the context, we improve the utilization of contextual information in the following three aspects. First, we design a novel context anchor to supervise high-level contextual feature learning by a semi-supervised method, which we call it PyramidAnchors. Second, we propose the Low-level Feature Pyramid Network to combine adequate high-level context semantic feature and Low-level facial feature together, which also allows the PyramidBox to predict faces of all scales in a single shot. Third, we introduce a contextsensitive structure to increase the capacity of prediction network to improve the final accuracy of output. In addition, we use the method of Data-anchor-sampling to augment the training samples across different scales, which increases the diversity of training data for smaller faces. By exploiting the value of context, PyramidBox achieves superior performance among the state-of-the-art over the two common face detection benchmarks, FDDB and WIDER FACE. Our code is available in Pad-dlePaddle: https://github.com/PaddlePaddle/models/tree/develop/ fluid/face_detection.
Recently, Bostan and his coauthors investigated lattice walks restricted to the non-negative octant N 3 . For the 35548 non-trivial models with at most six steps, they found that many models associated to a group of order at least 200 and conjectured these groups were in fact infinite groups. In this paper, we first confirm these conjectures and then consider the non-D-finite property of the generating function for some of these models.
The Visual Object Tracking challenge VOT2021 is the ninth annual tracker benchmarking activity organized by the VOT initiative. Results of 71 trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in recent years. The VOT2021 challenge was composed of four sub-challenges focusing on different tracking domains: (i) VOT-ST2021 challenge focused on short-term tracking in RGB, (ii) VOT-RT2021 challenge focused on "real-time" short-term tracking in RGB, (iii) VOT-LT2021 focused on long-term tracking, namely coping with target disappearance and reappearance and (iv) VOT-RGBD2021 challenge focused on long-term tracking in RGB and depth imagery. The VOT-ST2021 dataset was refreshed, while VOT-RGBD2021 introduces a training dataset and sequestered dataset for winner identification. The source code for most of the trackers, the datasets, the evaluation kit and the results along with the source code for most trackers are publicly available at the challenge website 1 .
Let p r (n) denote the number of r-component multipartitions of n, and let S γ,λ be the space spanned by η(24z) γ φ(24z), where η(z) is the Dedekind's eta function and φ(z) is a holomorphic modular form in M λ (SL 2 (Z)). In this paper, we show that the generating function of p r ( m k n+r 24 ) with respect to n is congruent to a function in the space S γ,λ modulo m k . As special cases, this relation leads to many well known congruences including the Ramanujan congruences of p(n) modulo 5, 7, 11 and Gandhi's congruences of p 2 (n) modulo 5 and p 8 (n) modulo 11. Furthermore, using the invariance property of S γ,λ under the Hecke operator T ℓ 2 , we obtain two classes of congruences pertaining to the m k -adic property of p r (n).
The robust feature extraction method for face representation is an important issue in face recognition. In this paper, we extract a new kind of feature through applying the idea of local binary pattern (LBP) into the resulted sub-images of Gabor transform. The new feature, i.e. Gabor-LBP-Like (GLLBP), together with its extension methods (1) overcome the drawback of losing information after Gabor transform’s down-sampling; (2) are insensitive to noise, compared with the LBP feature extracted from the original face image; and (3) are robust to image variation, especially occlusion and illumination changes when compared with other existing features combined LBP and Gabor transform. To validate the effectiveness of these features, we do experiments on the ORL, FERET, Georgia Tech and LFW facial databases. The numerical results show that GLLBP and its extensions are miraculous features for face recognition.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.