DPNeXt: A Lightweight Multi-Scale Feature Fusion Framework for Efficient ViT-Based Multi-Task Dense Prediction

SGP-Net: semantically-guided multi-task refinement for monocular depth estimation in scenes with potentially movable object categories

Semantically-guided progressive network (SGP-Net) is proposed, a semantically guided progressive refinement framework for MDE based on multi-task learning that improves key relative-error and accuracy metrics over the DCDepth baseline and remains competitive with recent methods.

BCSNet: Boundary-Centric Fusion and Selective Semantic Pyramid Injection Network for Real-Time Semantic Segmentation

Real-time semantic segmentation is a key component of resource-constrained perception systems, such as autonomous driving and robotic navigation, where dense scene understanding must be obtained under strict latency and computation constraints. Although dual-branch architectures provide an effective efficiency-oriented solution, they still suffer from boundary degradation during cross-branch fusion, insufficient multi-scale semantics for small and medium objects, and feature misalignment caused by content-agnostic upsampling near object contours. These limitations become more pronounced at high resolutions, where preserving thin structures and accurate class transitions is often in tension with maintaining high throughput. To address these issues, we present Boundary-Centric Fusion and Selective Semantic Pyramid Injection Network (BCSNet), a real-time segmentation framework that allocates lightweight modeling capacity to boundary-sensitive stages rather than increasing computation uniformly across the network. Specifically, the Boundary-Centric Cross-Branch Fusion and Refinement module learns a shared boundary cue to guide bidirectional feature exchange and local contour refinement with limited overhead. The Semantic Lightweight Feature Pyramid with Selective Injection module provides scale-adaptive semantic cues to the high-resolution stream through a compact pyramid design. The Boundary-Conditioned Region-Adaptive Alignment Upsampling operator further performs content-aware reassembly only within narrow boundary regions, while retaining efficient bilinear interpolation elsewhere. Under a controlled RTX 4090 evaluation protocol on Cityscapes, BCSNet-L achieves 79.5% mIoU at 107.0 FPS, while the lightweight BCSNet-S obtains 76.5% mIoU at 189.0 FPS. On CamVid, BCSNet achieves 77.1% mIoU at 156.8 FPS. These results indicate that BCSNet provides a practical accuracy--efficiency trade-off for high-resolution real-time segmentation, while direct embedded deployment and hardware-specific optimization remain directions for future work.

Improved DPT-Hybrid for Monocular Depth Estimation with Geometry-Enhanced Encoding and Structure-Aware Gated Fusion

Monocular depth estimation aims to recover dense 3D scene geometry from a single RGB image and plays an important role in autonomous driving, robotic perception, augmented reality, and 3D reconstruction. Although Transformer-based dense prediction models have achieved strong performance, existing DPT-Hybrid frameworks still suffer from three limitations: insufficient local geometric modeling in shallow stages, inadequate cross-scale fusion for preserving fine structures, and training objectives that only weakly constrain structural consistency. To address these issues, we propose a structure-aware enhanced DPT-Hybrid framework. First, a geometry-enhanced encoder introduces lightweight depth-wise separable convolution branches into shallow Transformer stages to better capture local edge and texture cues while preserving global contextual modeling. Second, a Structure-Aware Cross-Scale Gated Attention Fusion (S-GAF) module is proposed to improve decoder-side feature aggregation by jointly modeling channel-wise and spatial importance with an auxiliary RGB-gradient input. Third, joint structure–geometric consistency loss combines scale-invariant logarithmic loss, gradient consistency loss, and edge-focused loss to improve pixel-level accuracy, geometric plausibility, and boundary sharpness. Experiments on NYUv2 and KITTI demonstrate that the proposed method achieves lower single-run error metrics than the controlled DPT-Hybrid baseline under the evaluated settings. On NYUv2, our method achieves an absolute relative error (AbsRel) of 0.099 and an RMSE of 0.334. On KITTI, it achieves an AbsRel of 0.058 and an RMSE of 2.455. The proposed method introduces only modest additional complexity while producing more accurate and structurally sharper depth predictions.

ALLNet: Multi-task dense prediction for degraded images

Fusion-Oriented Deep Learning-Enhanced Visual SLAM: A Review

: Visual simultaneous localization and mapping (VSLAM) is a key technology for mobile robotics, autonomous driving, and embodied intelligence, enabling self-localization, environment reconstruction, and scene understanding. Although conventional geometric methods have achieved notable success, their performance often degrades in challenging conditions, such as low-texture scenes, severe illumination changes, dynamic interference, and long-term environmental variations. Recent advances in deep learning have created new opportunities to improve VSLAM through stronger feature representations, learned priors, semantic perception, and emerging map representations. At the same time, the increasing adoption of learning-based modules has raised important questions about integration strategies, generalization, interpretability, and real-time deployment. This paper presents a systematic review of deep learning-enhanced VSLAM, with a particular focus on how learning models are incorporated into classical simultaneous localization and mapping (SLAM) pipelines and how they function within the overall system. To provide a unified perspective, existing methods are organized into five categories according to their fusion interfaces with geometric SLAM pipelines: observation-level interfaces, constraint/prior/weight-level interfaces, solver-level interfaces, representation-level interfaces, and system-level integration interfaces. Based on this taxonomy, representative approaches are comparatively analyzed for accuracy, robustness, efficiency, and deployability. In addition, this review summarizes common design principles, including geometric consistency constraints, error propagation characteristics, and typical failure modes, and further discusses open challenges and future directions such as lightweight deployment, cross-domain adaptation, dynamic map modeling, and long-term consistency maintenance. This review aims to provide a structured reference for the analysis, design, and deployment of learning-enhanced VSLAM systems.

Enhancing 3D semantic scene completion via efficient attention and feature augmentation

3D Semantic Scene Completion (SSC), a cornerstone task in computer vision, aims to simultaneously predict the geometric structure and semantic content of 3D scenes from sparse inputs. This capability is crucial for advancing applications in robotics, autonomous driving, and virtual reality. However, SSC faces significant challenges, including the high computational cost of capturing long-range contextual information and the scarcity of 3D semantic labels leading to overfitting. To address these limitations, we propose an enhanced network for semantic scene completion. Firstly, we devise a 3D Local- Global Linear Attention Mechanism (LG-LAM) that efficiently captures long-range contextual information with linear complexity, enabling a comprehensive understanding of the 3D scene without heavy computational burdens. Secondly, a 3D Feature Augmentation Module (FAM) is integrated to enrich feature diversity through rotation-invariant learning, mitigating overfitting and enhancing the model’s robustness given limited annotations. Extensive experiments on the NYUCAD dataset demonstrate that our method achieves state-of-the-art performance among non-iterative methods while introducing negligible computational overhead.