Search

We address the issue of 3D face modeling and tracking from RGB-D images. Existing methods usually fit a deformable model to an RGB-D image using iterative closest point algorithm. Due to the noise and occlusion of the depth image, these methods are not robust enough. To solve this issue, we propose a method for robust 3D face modeling and tracking. For an input RGB-D face image, our method first estimates the initial head pose of a person using random forests. Then, a generic bilinear face model is fitted to the RGB-D image using iterative closest point algorithm. To improve the accuracy and robustness of face modeling, an optimal weight for each face vertex is integrated into the fitting procedure. The distances between facial landmarks are also used to better estimate facial expressions. Finally, the head pose, the identity, and expression parameters of the bilinear face model are jointly optimized. Experiments show that our method can generate accurate 3D face models from an RGB-D image or image sequence. The method can also robustly track the face even if the person is with large head rotations and various facial expressions. [ABSTRACT FROM AUTHOR]

The long short-term memory (LSTM) network with gating mechanism has been widely used in sequence modeling tasks including handwriting and speech recognition. As an LSTM network can be unfolded along the temporal dimension and its temporal depth is equal to the length of the input feature sequence, the introduction of gating might not be sufficient to completely model the dynamic temporal dependencies in sequential data. Inspired by the residual learning in ResNet, this paper proposes a dynamic temporal residual network (DTRN) by incorporating residual learning into an LSTM network along the temporal dimension. DTRN involves two networks: Its primary network consists of modified LSTM units with weighted shortcut connections for adjacent temporal outputs, while its secondary network generates dynamic weights for the shortcut connections. To validate the performance of DTRN, we conduct experiments on three commonly used public handwriting recognition datasets (IFN/ENIT, IAM and Rimes) and one speech recognition dataset (TIMIT). The experimental results show that the proposed DTRN has outperformed previously reported methods. [ABSTRACT FROM AUTHOR]

This paper considers the task of image search using the Bag-of-Words (BoW) model. In this model, the precision of visual matching plays a critical role. Conventionally, local cues of a keypoint, e.g., SIFT, are employed. However, such strategy does not consider the contextual evidences of a keypoint, a problem which would lead to the prevalence of false matches. To address this problem and enable accurate visual matching, this paper proposes to integrate discriminative cues from multiple contextual levels, i.e., local, regional, and global, via probabilistic analysis. 'True match' is defined as a pair of keypoints corresponding to the same scene location on all three levels (Fig. 1). Specifically, the Convolutional Neural Network (CNN) is employed to extract features from regional and global patches. We show that CNN feature is complementary to SIFT due to its semantic awareness and compares favorably to several other descriptors such as GIST, HSV, etc. To reduce memory usage, we propose to index CNN features outside the inverted file, communicated by memory-efficient pointers. Experiments on three benchmark datasets demonstrate that our method greatly promotes the search accuracy when CNN feature is integrated. We show that our method is efficient in terms of time cost compared with the BoW baseline, and yields competitive accuracy with the state-of-the-arts. [ABSTRACT FROM AUTHOR]

Recently, the bag-of-words representation is widely applied in the image retrieval applications. In this model, visual word is a core component. However, compared with text retrieval, one major problem associated with image retrieval consists in the visual word ambiguity, i.e., a trade-off between precision and recall of visual matching. To address this problem, this paper proposes a tensor index structure to improve precision and recall simultaneously. Essentially, the tensor index is a multi-dimensional index structure. It combines the strengths of two state-of-the-art indexing strategies, i.e., the inverted multi-index [Babenko and Lempitsky (Computer vision and pattern recognition (CVPR), 2012 IEEE Conference, 3069-3076, )] as well as the joint inverted index [Xia et al. (ICCV, )] which are initially designed for approximate nearest neighbor search problems. This paper, instead, exploits their usage in the scenario of image retrieval and provides insights into how to combine them effectively. We show that on the one hand, the multi-index enhances the discriminative power of visual words, thus improving precision; on the other hand, the introduction of multiple codebooks corrects quantization artifacts, thus improving recall. Extensive experiments on two benchmark datasets demonstrate that tensor index significantly improves the baseline approach. Moreover, when incorporating methods such as Hamming embedding, we achieve competitive performances compared to the state-of-the-art ones. [ABSTRACT FROM AUTHOR]

By gel casting with polymethylmethacrylate microbeads (PMMA) as pore-forming agent, porous boron nitride/silicon nitride (BN/SiN) composite ceramics were successfully prepared. The obtained ceramic shows bimodal hierarchical structures that composed of spherical-shaped micro pores depending on PMMA content and irregular sub-micro pores formed by the stacking of ceramic particles. Porosity of the porous BN/SiN ceramics can be well controlled from 53.0 to 60.6 % by the PMMA content from 10 to 40 wt%, as well as the mechanical and dielectric properties. Effect of PMMA content on phase composition and the relationship between microstructure and the basic properties of the porous BN/SiN ceramics was discussed in detail. Microstructure analysis reveals that the sub-micro pores acted as channels between micro pores. BN particles have a relatively denser distribution on the wall of spherical-shaped micro pores with a window between micro and sub-micro pores, and resulting in a half-closed micro pore structure, which is meaningful for material design with concentration of BN particles on the wall of pore structure. [ABSTRACT FROM AUTHOR]