Semantic segmentation is a fundamental task in computer vision with wide-ranging applications, including autonomous driving and robotics. While RGB-based methods have achieved strong performance with CNNs and Transformers, their effectiveness degrades under fast motion, low-light, or high dynamic range conditions due to limitations of frame cameras. Event cameras offer complementary advantages such as high temporal resolution and low latency, yet lack color and texture, making them insufficient on their own. To address this, recent research has explored multimodal fusion of RGB and event data; however, many existing approaches are computationally expensive and focus primarily on spatial fusion, neglecting the temporal dynamics inherent in event streams. In this work, we propose MambaSeg, a novel dual-branch semantic segmentation framework that employs parallel Mamba encoders to efficiently model RGB images and event streams. To reduce cross-modal ambiguity, we introduce the Dual-Dimensional Interaction Module (DDIM), comprising a Cross-Spatial Interaction Module (CSIM) and a Cross-Temporal Interaction Module (CTIM), which jointly perform fine-grained fusion along both spatial and temporal dimensions. This design improves cross-modal alignment, reduces ambiguity, and leverages the complementary properties of each modality. Extensive experiments on the DDD17 and DSEC datasets demonstrate that MambaSeg achieves state-of-the-art segmentation performance while significantly reducing computational cost, showcasing its promise for efficient, scalable, and robust multimodal perception.

Unmanned Aerial Vehicles (UAVs) play an important role in various applications, where precise trajectory tracking is crucial. However, conventional control algorithms for trajectory tracking often exhibit limited performance due to the underactuated, nonlinear, and highly coupled dynamics of quadrotor systems. To address these challenges, we propose HBO-PID, a novel control algorithm that integrates the Heteroscedastic Bayesian Optimization (HBO) framework with the classical PID controller to achieve accurate and robust trajectory tracking. By explicitly modeling input-dependent noise variance, the proposed method can better adapt to dynamic and complex environments, and therefore improve the accuracy and robustness of trajectory tracking. To accelerate the convergence of optimization, we adopt a two-stage optimization strategy that allow us to more efficiently find the optimal controller parameters. Through experiments in both simulation and real-world scenarios, we demonstrate that the proposed method significantly outperforms state-of-the-art (SOTA) methods. Compared to SOTA methods, it improves the position accuracy by 24.7% to 42.9%, and the angular accuracy by 40.9% to 78.4%.

Event-based semantic segmentation has great potential in autonomous driving and robotics due to the advantages of event cameras, such as high dynamic range, low latency, and low power cost. Unfortunately, current artificial neural network (ANN)-based segmentation methods suffer from high computational demands, the requirements for image frames, and massive energy consumption, limiting their efficiency and application on resource-constrained edge/mobile platforms. To address these problems, we introduce SLTNet, a Spike-driven Lightweight Transformer-based Network designed for event-based semantic segmentation. Specifically, SLTNet is built on efficient spike-driven convolution blocks (SCBs) to extract rich semantic features while reducing the model’s parameters. Then, to enhance the long-range contextual feature interaction, we propose novel spike-driven transformer blocks (STBs) with binary mask operations. Based on these basic blocks, SLTNet employs a high-efficiency single-branch architecture while maintaining the low energy consumption of the Spiking Neural Network (SNN). Finally, extensive experiments on DDD17 and DSEC-Semantic datasets demonstrate that SLTNet outperforms state-of-the-art (SOTA) SNN-based methods by at most 9.06% and 9.39% mIoU, respectively, with extremely 4.58× lower energy consumption and 114 FPS inference speed. Our code is open-sourced and available at https://github.com/longxianlei/SLTNet-v1.0.

Event-based semantic segmentation in traffic scenes has attracted considerable attention in autonomous driving systems due to the advantages of event cameras such as high dynamic range, low latency, and low energy consumption. However, existing Artificial Neural Network (ANN)-based methods rely on conventional image frames, often neglecting the spatial-temporal dynamics inherent in event streams and consuming higher energy costs, significantly limiting their applicability in energy-constrained environments. In this study, we introduce Spike-BRGNet, a Spike-driven Boundary Region-Guided Network that efficiently extracts boundary information utilizing only events to guide the segmentation encoder, while preserving the energy efficiency of Spiking Neural Networks (SNNs). Specifically, to explore the implicit information from events, we design a three-branch spiking encoder that consists of semantic detail (SD), context aggregation (CA), and boundary aware (BA) branches to capture specific features. Then, a spiking multi-scale context aggregation (SMSCA) module is proposed to enhance the semantics of the CA branch. Finally, a novel boundary region-guided loss function and a dynamic surrogate gradient function, EvAF, are designed to optimize the model. Extensive experiments show that our model outperforms state-of-the-art (SOTA) SNN-based methods on DDD17 (+1.57%) and DSEC dataset (+1.91%). Furthermore, Spike-BRGNet consumes$17.76\times $less energy than ANN-based models, showing superior energy-saving performance.

Visual Simultaneous Localization and Mapping (SLAM) is crucial to many applications such as self-driving vehicles and robot tasks. However, it is still challenging for existing visual SLAM approaches to achieve good performance in low-texture or illumination-changing scenes. In recent years, some researchers have turned to edge-based SLAM approaches to deal with the challenging scenes, which are more robust than feature-based and direct SLAM methods. Nevertheless, existing edge-based methods are computationally expensive and inferior than other visual SLAM systems in terms of accuracy. In this study, we propose EdgeSLAM, a novel RGB-D edge-based SLAM approach to deal with challenging scenarios that is efficient, accurate, and robust. EdgeSLAM is built on two innovative modules: efficient edge selection and adaptive robust motion estimation. The edge selection module can efficiently select a small set of edge pixels, which significantly improves the computational efficiency without sacrificing the accuracy. The motion estimation module improves the system's accuracy and robustness by adaptively handling outliers in motion estimation. Extensive experiments were conducted on TUM RGBD, ICL-NUIM and ETH3D datasets, and experimental results show that EdgeSLAM significantly outperforms five state-of-the-art (SOTA) methods in terms of efficiency, accuracy, and robustness, which achieves 29.17% accuracy improvements with a high processing speed of up to 120 FPS and a high positioning success rate of 97.06%.

In recent years, wide-area visual surveillance systems have been widely applied in various industrial and transportation scenarios. These systems, however, face significant challenges when implementing multi-object detection due to conflicts arising from the need for high-resolution imaging, efficient object searching, and accurate localization. To address these challenges, this paper presents a hybrid system that incorporates a wide-angle camera, a high-speed search camera, and a galvano-mirror. In this system, the wide-angle camera offers panoramic images as prior information, which helps the search camera capture detailed images of the targeted objects. This integrated approach enhances the overall efficiency and effectiveness of wide-area visual detection systems. Specifically, in this study, we introduce a wide-angle camera-based method to generate a panoramic probability map (PPM) for estimating high-probability regions of target object presence. Then, we propose a probability searching module that uses the PPM-generated prior information to dynamically adjust the sampling range and refine target coordinates based on uncertainty variance computed by the object detector. Finally, the integration of PPM and the probability searching module yields an efficient hybrid vision system capable of achieving 120 fps multi-object search and detection. Extensive experiments are conducted to verify the system’s effectiveness and robustness.

With the increasing demand for high precision and low power consumption in Human Activity Recognition (HAR) techniques, deep learning-based HAR models have emerged as the hottest research topics. Due to the excellent feature extraction and modeling abilities of deep learning models, which enable them to fit a wide variety of complex patterns. However, these models often require a large number of parameters, leading to high computational costs and longer processing time. These inherent factors pose significant challenges for resource-constraint edge devices to perform efficient HAR. To address these issues, we propose MobileHAR, which combines depthwise separable convolutions and novel Inverted Residual Inception Blocks (IRIB). This combination significantly reduces computational load and frequent memory access while maintaining high recognition accuracy. Then, we design a special class imbalance loss to supervise the model to pay more attention to imbalance classes. Finally, extensive experiments on several public datasets demonstrate that our method improves accuracy by 3.15% compared to traditional methods and requires only 0.15M parameters, which is at least four times fewer than the compared methods.

Sensor-rich smartphones have facilitated a lot of services and applications. Indoor/Outdoor (IO) status serves as a critical foundation for various upstream tasks, including seamless pedestrian navigation, power management, and activity recognition. Nevertheless, achieving robust, efficient, and accurate IO detection remains challenging due to environmental complexities and device heterogeneity. To tackle this challenge, some researchers have turned to deep learning for IO detection, which can deal with complex scenarios and achieve high detection accuracy. However, deep learning methods are often blamed for their expensive computational cost. Therefore, in this paper, we introduce a novel efficient IO detection method-DeepSIO, which can detect IO status accurately and efficiently. Specifically, different from existing IO detection methods, DeepSIO is developed based on spiking neural networks (SNN) that are more biologically plausible and computationally efficient than other deep neural networks. To better capture useful features, we propose to utilize dense connections between SNN layers. Extensive experiments are conducted in three typical scenarios, and experimental results demonstrate that DeepSIO outperforms state-of-the-art methods, achieving an accuracy of about 99.7%. Moreover, it has better generalization ability and can adapt well to new environments and devices.

As an external manifestation of human emotions, expression recognition plays an important role in human-computer interaction. Although existing expression recognition methods performs perfectly on constrained frontal faces, there are still many challenges in expression recognition in natural scenes due to different unrestricted conditions. Expression classification belongs to a pattern recognition problem where intra-class distance is greater than the inter-class distance, which leads to severe over-fitting when using neural networks for expression recognition. This paper proposes a novel net-work structure called Dimension Reduction Network which can effectively reduce generalization error. By adding a data dimension reduction module before the general classification network, a lot of redundant information is filtered, and only useful information is left. This can reduce the interference by irrelevant information when performing classification tasks and reduce generalization error. The proposed method does not require any modification to the classification network, only a small dimension reduction module needs to be added in front of the classification network. However, it can effectively reduce generalization error. We designed big and tiny versions of Dimension Reduction Network, both exceeds our baseline on AffectNet data set. The big version of our proposed method surpassed the state-of-the-art methods by more than 1.2% on AffectNet data set. Our code will open source3when the paper is accepted.

Deep neural networks (DNNs) have achieved great success in a wide range of computer vision areas, but the applications to mobile devices is limited due to their high storage and computational cost. Much efforts have been devoted to compress DNNs. In this paper, we propose a simple yet effective method for deep networks compression, named Cluster Regularized Quantization (CRQ), which can reduce the presentation precision of a full-precision model to ternary values without significant accuracy drop. In particular, the proposed method aims at reducing the quantization error by introducing a cluster regularization term, which is imposed on the full-precision weights to enable them naturally concentrate around the target values. Through explicitly regularizing the weights during the re-training stage, the full-precision model can achieve the smooth transition to the low-bit one. Comprehensive experiments on benchmark datasets demonstrate the effectiveness of the proposed method.