In practical Internet of Things (IoT) applications, deep neural networks (DNNs) often encounter challenges arising from covariate shifts (differences in feature distributions) and category shifts (discrepancies in label spaces), which significantly degrade their generalization performance. To mitigate these issues, universal federated domain adaptation (UFDA) techniques have been proposed to train a global model that can classify known and unknown categories while keeping data private. Nevertheless, most existing methods still struggle to precisely identify samples belonging to unknown classes in the target domain due to the unavailability of data from the source domain clients. To address these challenges, we propose a novel method, termed one-vs-all self-supervision (OSS) for IoT scenario. Specifically, OSS mainly consists of following three components. First, one-vs-all pseudo-label generation is proposed to generate high-quality pseudo-labels by leveraging source client models. Subsequently, we design a category-diverse strategy to aggregate the source models by assigning appropriate weights to each source domain client. Finally, we implement a target self-supervised learning strategy to refine feature alignment with respect to cluster centers. Comprehensive experiments are performed on four benchmark datasets: Office-31, Office-Home, VisDA-2017+ImageCLEF-DA, and Digits. The results show that our proposed OSS method achieves state-of-the-art performance in UFDA, significantly enhancing the recognition accuracy.

Source-Free Object Detection (SFOD) enables knowledge transfer from a source domain to an unsupervised target domain for object detection without access to source data. Most existing SFOD approaches are either confined to conventional object detection (OD) models like Faster R-CNN or designed as general solutions without tailored adaptations for novel OD architectures, especially Detection Transformer (DETR). In this paper, we introduce Feature Reweighting ANd Contrastive Learning NetworK (FRANCK), a novel SFOD framework specifically designed to perform query-centric feature enhancement for DETRs. FRANCK comprises four key components: 1) an Objectness Score-based Sample Reweighting (OSSR) module that computes attention-based objectness scores on multi-scale encoder feature maps, reweighting the detection loss to emphasize less-recognized regions; 2) a Contrastive Learning with Matching-based Memory Bank (CMMB) module that integrates multi-level features into memory banks, enhancing class-wise contrastive learning; 3) an Uncertainty-weighted Query-fused Feature Distillation (UQFD) module that improves feature distillation through prediction quality reweighting and query feature fusion; and 4) an improved self-training pipeline with a Dynamic Teacher Updating Interval (DTUI) that optimizes pseudo-label quality. By leveraging these components, FRANCK effectively adapts a source-pre-trained DETR model to a target domain with enhanced robustness and generalization. Extensive experiments on several widely used benchmarks demonstrate that our method achieves state-of-the-art performance, highlighting its effectiveness and compatibility with DETR-based SFOD models.

Despite recent advances in lane detection methods, scenarios with limited- or no-visual-clue of lanes due to factors such as lighting conditions and occlusion remain challenging and crucial for automated driving. Moreover, current lane representations require complex post-processing and struggle with specific instances. Inspired by the DETR architecture, we propose LDTR, a transformer-based model to address these issues. Lanes are modeled with a novel anchor-chain, regarding a lane as a whole from the beginning, which enables LDTR to handle special lanes inherently. To enhance lane instance perception, LDTR incorporates a novel multi-referenced deformable attention module to distribute attention around the object. Additionally, LDTR incorporates two line IoU algorithms to improve convergence efficiency and employs a Gaussian heatmap auxiliary branch to enhance model representation capability during training. To evaluate lane detection models, we rely on Fréchet distance, parameterized Fl-score, and additional synthetic metrics. Experimental results demonstrate that LDTR achieves state-of-the-art performance on well-known datasets.

Lane detection is challenging due to the complicated onroad scenarios and line deformation from different camera perspectives. Lots of solutions were proposed, but can not deal with "corner lanes" well. To address this problem, this paper proposes a new top-down deep learning lane detection approach, CANet. A lane instance is first responded by the heatmap on the U-shaped "curved guide line" at global semantic level, thus the corresponding features of each lane are aggregated at the response point. Then CANet obtains the heatmap response of the entire lane through conditional convolution, and finally decodes the point set to describe lanes via adaptive decoder. The prototype is implemented with Pytorch, and evaluated against 3 well-known datasets extensively. The experimental results show that CANet reaches SOTA in different metrics.

Semi-supervised semantic segmentation methods are the main solution to alleviate the problem of high annotation consumption in semantic segmentation. However, the class imbalance problem makes the model favor the head classes with sufficient training samples, resulting in poor performance of the tail classes. To address this issue, we propose a Decoupled Semi-Supervise Semantic Segmentation (DeS4) framework based on the teacher-student model. Specifically, we first propose a decoupling training strategy to split the training of the encoder and segmentation decoder, aiming at a balanced decoder. Then, a non-learnable prototype-based segmentation head is proposed to regularize the category representation distribution consistency and perform a better connection between the teacher model and the student model. Furthermore, a Multi-Entropy Sampling (MES) strategy is proposed to collect pixel representation for updating the shared prototype to get a class-unbiased head. We conduct extensive experiments of the proposed DeS4 on two challenging benchmarks (PASCAL VOC 2012 and Cityscapes) and achieve remarkable improvements over the previous state-of-the-art methods.

Emotion recognition in user-generated videos plays an important role in human-centered computing. Existing methods mainly employ traditional two-stage shallow pipeline, i.e. extracting visual and/or audio features and training classifiers. In this paper, we propose to recognize video emotions in an end-to-end manner based on convolutional neural networks (CNNs). Specifically, we develop a deep Visual-Audio Attention Network (VAANet), a novel architecture that integrates spatial, channel-wise, and temporal attentions into a visual 3D CNN and temporal attentions into an audio 2D CNN. Further, we design a special classification loss, i.e. polarity-consistent cross-entropy loss, based on the polarity-emotion hierarchy constraint to guide the attention generation. Extensive experiments conducted on the challenging VideoEmotion-8 and Ekman-6 datasets demonstrate that the proposed VAANet outperforms the state-of-the-art approaches for video emotion recognition. Our source code is released at: https://github.com/maysonma/VAANet.

Although promising results have been achieved in the areas of object detection and classification, few works have provided an end-to-end solution to the perception problems in the autonomous driving field. In this paper, we make two contributions. Firstly, we fully enhanced our previously released TT100K benchmark and provide 16,817 elaborately labeled Tencent Street View panoramas. This newly created benchmark, we call it Tencent Autonomous Driving 16K (TAD16K), not only contains previously labeled traffic-signs (221 types), but also creates annotations for three new objects, which are traffic lights (6 types), vehicles and pedestrians. Secondly, we provide the evaluation results of two state-of-the-art object detection algorithms (SSD and DetectNet) on our benchmark, which can be used as the baseline for future comparison purpose. Finally, we also demonstrate that the network trained on our benchmark can be directly deployed for practical application. The TAD16K, relevant additions and the source codes are publicly available1,2.

In this paper, we focus on subcarrier and power allocation for layered multicast streams in OFDMA cellular networks, where the multicast stream is composed of a basic layer and an enhancement layer. Our goal is to maximize the system total throughput with a total power constraint and a minimum rate requirement. A low-complexity allocation algorithm is proposed, which combined the suitability-based subcarrier allocation (SSA) with the traditional water filling (TWF) for the base layer and an advanced water filling (AWF) for the enhancement layer. Besides, we present a throughput-based user selection (TUS) algorithm which selects a proper set of users to serve when the minimum rate requirement can not be satisfied. Simulation results show that our proposed algorithm can improve the system throughput and outage probability.

In this paper, a 4-transmit-antenna space time block coding (STBC) scheme with 1-bit differential feedback is proposed. This scheme gives full transmit diversity and spatial coding rate over time-selective fading channels by rotating the signal constellations for certain angles determined by the differential feedback information. By utilizing differential feedback, our method can provide much better channel orthogonality and track the variation of the channel more accurately and timely under time-selective fading circumstances, comparing to recent works presenting other quantized feedback schemes. Simulation results show that the 1-bit differential feedback scheme achieves near-optimum performance and outperforms existing algorithms without adding more overhead or complexity.

An improved design of receiving filter for non- ideally shaped modulated signals which provides better performance than existing schemes is proposed. By concerning both Inter-Symbol-Interference (ISI) and the degree of waveform mismatch between the transmitted signal and impulse response of the receiving filter, the exact expression of the Signal-to-Noise Ratio (SNR) loss that represents the performance degradation is derived, and the performance is compared with that of the ideally shaped signals' demodulation. The existing schemes such as root-raised-cosine (RRC) receiving filters don't perform well for non-ideally shaped situations, since large degrees of both ISI and waveform mismatch exist. To improve the performance and avoid the complexity of applying equalizer in receivers, the proposed scheme designs the impulse response of the receiving filter by optimizing the SNR loss to the minimum value, and it uses simulated annealing as the optimization algorithm. It is shown that the new method can achieve better performance than existing schemes, especially when the modulation order or the SNR is relatively high. Finally the conclusion is validated by simulation results.

Signal-to-noise ratio (SNR) estimation is an important issue in many communication systems. This paper proposed a new non-data-aided (NDA) SNR estimator for BPSK and QPSK modulated signals. The estimation is based on pseudo-error probability (PEP), which can be estimated without additional cost for sampled systems. The performance of the proposed estimator is examined numerically in terms of its bias and normalized mean square error. The Cramer-Rao lower bound (CRLB) for PEP-based SNR estimation is also derived. Comparing with the existing SNR estimation schemes such as maximum likelihood (ML) estimators, the proposed algorithm can achieve similar performance with much less computational cost.