As a significant biometric identification technology, face recognition (FR) is extensively utilized in identity verification and security surveillance systems. Current research predominantly relies on high-definition RGB camera-based methods. However, these methods are susceptible to various factors such as lighting conditions and disguises. This paper proposes a low-cost face recognition solution called Warm- Face based on thermal array sensors. By leveraging the thermal radiation of the face, we overcome the disturbances caused by lighting conditions and disguises, thereby achieving rapid and highly accurate face recognition. However, face recognition based on thermal array sensors still faces two major challenges. Firstly, in complex scenarios, thermal noise interference can lead to the thermal radiation characteristics of the target face becoming indistinguishable from the background. Secondly, due to their large network parameter sizes and high computational complexity, recognition models face challenges in simultaneously achieving low latency and high accuracy. WarmFace extracts facial regions through a semantic segmentation-based approach, effectively reducing the impact of background interference on recognition performance. Additionally, in the recognition model, we utilize a series of linear transformations instead of convolution operations to process the intrinsic features of images, which reduces redundancy in feature maps while preserving the essential information. Extensive real-world experiments validate the effectiveness of WarmFace in various environments, achieving an average recognition accuracy of 98.6%.

Person re-identification (Re-ID) can recognize users based on their clothing, body shape, and other information without the need for clear facial images, and is widely applied in the field of intelligent security. Traditional Re-ID systems mainly rely on high-definition RGB cameras, but the deployment of large-scale high-definition RGB cameras indoors has caused serious privacy and ethical concerns. Recently, wireless-based Re-ID systems (Wi-Fi, RFID, millimeter-wave radar, etc.) have shown promising prospects, but the limited sensing resolution hinders their practical deployment. In this paper, we propose WarmGait, a Re-ID system based on thermal array sensors, which can achieve high-precision Re-ID at low cost and minimize the invasion of user privacy. However, using thermal arrays for Re-ID still faces two major challenges. The first is the low and unclear texture resolution of images caused by low-cost infrared devices. The second is that existing gait recognition methods require maintaining the sequential constraint of gait images, which reduces the flexibility of gait recognition or Re-ID. To address these two challenges, we first designed an edge module inspired by Taylor Finite Difference (TFD) to aggregate image edge information to help improve the resolution of infrared devices. Then, we considered gait as a collection of gait profiles and extracted features from the frame level and collection level for recognition, breaking through the limitations of the number and order of input images. After extensive experimental evaluation, our model can achieve an average recognition accuracy of 87.3% in various scenarios, demonstrating the potential of WarmGait in Re-ID.

Convolution operations form the computational backbone of deep learning inference but often become performance bottlenecks on conventional architectures. While multi-core Digital Signal Processors (DSPs) offer energy-efficient alternatives through long vector units and software-managed memory hierarchies, existing convolution optimizations designed for CPUs/GPUs underperform due to architectural mismatches in memory systems and execution pipelines. We present mtConv, an optimizing convolution method for multi-core DSPs. mtConv achieves high performance by exploiting data reuse, managing on-chip memory, and designing efficient micro-kernels. It maximizes the overlap between computation and communication to hide data transfer latency, leveraging DSPs’ long vector units and hierarchical scratchpad memories. We evaluate mtConv against state-of-the-art convolution optimizations on DSPs. Experimental results show that mtConv delivers the best overall performance across various convolution layers, achieving up to 93.25% of the hardware’s peak performance on a single DSP core and 92.31% when using all 8 cores of a single DSP cluster.

Human can infer the 3D geometry of a scene from a sketch instead of a realistic image, which indicates that the spatial structure plays a fundamental role in understanding the depth of scenes. We are the first to explore the learning of a depth-specific structural representation, which captures the essential feature for depth estimation and ignores irrelevant style information. Our S2R-DepthNet (Synthetic to Real DepthNet) can be well generalized to un-seen real-world data directly even though it is only trained on synthetic data. S2R-DepthNet consists of: a) a Structure Extraction (STE) module which extracts a domain-invariant structural representation from an image by dis-entangling the image into domain-invariant structure and domain-specific style components, b) a Depth-specific Attention (DSA) module, which learns task-specific knowledge to suppress depth-irrelevant structures for better depth estimation and generalization, and c) a depth prediction module (DP) to predict depth from the depth-specific representation. Without access of any real-world images, our method even outperforms the state-of-the-art unsupervised domain adaptation methods which use real-world images of the tar-get domain for training. In addition, when using a small amount of labeled real-world data, we achieve the state-of-the-art performance under the semi-supervised setting.

As a result of collisions of signaling packets when a large number of terminals access the PDT system randomly, a large number of PDT terminals frequently register slowly and some PDT terminals are not successfully registered. In order to solve the above problem, this paper studies the main factors that affect the registration time, including retransmission probability and the number of terminals. Firstly, the random access technology based on slotted ALOHA is introduced, and to the collision delay is analyzed. Secondly, Markov chain is used for mathematical modeling to provide theoretical basis of simulation research. Finally, the influences of retransmission probability and terminal number on terminal registration time are analyzed by MATLAB simulation software, which provides guidance for engineering practice.

Abstract We collected research articles from Retraction Watch database, Scopus, and a major retraction announcement by Springer, to identify emails used by authors. Authors' emails can be institutional emails and noninstitutional emails. Data suggest that retracted articles are more likely to use noninstitutional emails, but it is difficult to generalize. The study put some focus on authors from China.

Many computer vision tasks, such as monocular depth estimation and height estimation from a satellite orthophoto, have a common underlying goal, which is regression of dense continuous values for the pixels given a single image. We define them as dense continuous-value regression (DCR) tasks. Recent approaches based on deep convolutional neural networks significantly improve the performance of DCR tasks, particularly on pixelwise regression accuracy. However, it still remains challenging to simultaneously preserve the global structure and fine object details in complex scenes. In this article, we take advantage of the efficiency of Laplacian pyramid on representing multiscale contents to reconstruct high-quality signals for complex scenes. We design a Laplacian pyramid neural network (LAPNet), which consists of a Laplacian pyramid decoder (LPD) for signal reconstruction and an adaptive dense feature fusion (ADFF) module to fuse features from the input image. More specifically, we build an LPD to effectively express both global and local scene structures. In our LPD, the upper and lower levels, respectively, represent scene layouts and shape details. We introduce a residual refinement module to progressively complement high-frequency details for signal prediction at each level. To recover the signals at each individual level in the pyramid, an ADFF module is proposed to adaptively fuse multiscale image features for accurate prediction. We conduct comprehensive experiments to evaluate a number of variants of our model on three important DCR tasks, i.e., monocular depth estimation, single-image height estimation, and density map estimation for crowd counting. Experiments demonstrate that our method achieves new state-of-the-art performance in both qualitative and quantitative evaluation on the NYU-D V2 and KITTI for monocular depth estimation, the challenging Urban Semantic 3D (US3D) for satellite height estimation, and four challenging benchmarks for crowd counting. These results demonstrate that the proposed LAPNet is a universal and effective architecture for DCR problems.

Monocular depth estimation is an essential task for scene understanding. The underlying structure of objects and stuff in a complex scene is critical to recovering accurate and visually-pleasing depth maps. Global structure conveys scene layouts, while local structure reflects shape details. Recently developed approaches based on convolutional neural networks (CNNs) significantly improve the performance of depth estimation. However, few of them take into account multi-scale structures in complex scenes. In this paper, we propose a Structure-Aware Residual Pyramid Network (SARPN) to exploit multi-scale structures for accurate depth prediction. We propose a Residual Pyramid Decoder (RPD) which expresses global scene structure in upper levels to represent layouts, and local structure in lower levels to present shape details. At each level, we propose Residual Refinement Modules (RRM) that predict residual maps to progressively add finer structures on the coarser structure predicted at the upper level. In order to fully exploit multi-scale image features, an Adaptive Dense Feature Fusion (ADFF) module, which adaptively fuses effective features from all scales for inferring structures of each scale, is introduced. Experiment results on the challenging NYU-Depth v2 dataset demonstrate that our proposed approach achieves state-of-the-art performance in both qualitative and quantitative evaluation. The code is available at https://github.com/Xt-Chen/SARPN.