We present HI-SLAM2, a geometry-aware Gaussian SLAM system that achieves fast and accurate monocular scene reconstruction using only RGB input. Existing Neural SLAM or 3DGS-based SLAM methods often trade off between rendering quality and geometry accuracy, our research demonstrates that both can be achieved simultaneously with RGB input alone. The key idea of our approach is to enhance the ability for geometry estimation by combining easy-to-obtain monocular priors with learning-based dense SLAM, and then using 3D Gaussian splatting as our core map representation to efficiently model the scene. Upon loop closure, our method ensures on-the-fly global consistency through efficient pose graph bundle adjustment and instant map updates by explicitly deforming the 3D Gaussian units based on anchored keyframe updates. Furthermore, we introduce a grid-based scale alignment strategy to maintain improved scale consistency in prior depths for finer depth details. Through extensive experiments on Replica, ScanNet, and ScanNet++, we demonstrate significant improvements over existing Neural SLAM methods and even surpass RGB-D-based methods in both reconstruction and rendering quality.

Hongyi Luo, Qing Cheng, Daniel Matos, Hari Krishna Gadi, Yanfeng Zhang, Lu Liu, Yongliang Wang, Niclas Zeller, Daniel Cremers, Liqiu Meng. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Radar-based object detection has become a popular research topic in the past few years because of the resilience of radar sensors in adverse lighting and weather conditions compared to traditional camera and LiDAR systems. Most radar deep learning models utilize a processed form of the radar signal as it is easier to extract essential information from these forms. However, these processing steps can significantly increase the inference time and ultimately limit the ability to use radar object detection in a real-time environment on autonomous vehicles. Recent research has been able to utilize raw radar signals to maintain fast inference speeds, but this has come at the cost of performance. We build on this research by first introducing the Separable Convolution (SCON) block, a lightweight CNN-based block to extract local features. The SCON block is combined with the global feature extraction capabilities of Swin Transformers to create a highly efficient backbone. This hybrid CNN-Swin backbone was used for object detection and classification on the RADDet dataset and achieved a mAP of 55.6%, which is state-of-the-art for raw radar input, and achieves state-of-the-art parity for processed radar inputs. These results were achieved while also maintaining a fast inference time.

Abstract In this paper, we present a novel visual SLAM and long-term localization benchmark for autonomous driving in challenging conditions based on the large-scale 4Seasons dataset. The proposed benchmark provides drastic appearance variations caused by seasonal changes and diverse weather and illumination conditions. While significant progress has been made in advancing visual SLAM on small-scale datasets with similar conditions, there is still a lack of unified benchmarks representative of real-world scenarios for autonomous driving. We introduce a new unified benchmark for jointly evaluating visual odometry, global place recognition, and map-based visual localization performance which is crucial to successfully enable autonomous driving in any condition. The data has been collected for more than one year, resulting in more than 300 km of recordings in nine different environments ranging from a multi-level parking garage to urban (including tunnels) to countryside and highway. We provide globally consistent reference poses with up to centimeter-level accuracy obtained from the fusion of direct stereo-inertial odometry with RTK GNSS. We evaluate the performance of several state-of-the-art visual odometry and visual localization baseline approaches on the benchmark and analyze their properties. The experimental results provide new insights into current approaches and show promising potential for future research. Our benchmark and evaluation protocols will be available at https://go.vision.in.tum.de/4seasons .

We present HI-SLAM2, a geometry-aware Gaussian SLAM system that achieves fast and accurate monocular scene reconstruction using only RGB input. Existing Neural SLAM or 3DGS-based SLAM methods often trade off between rendering quality and geometry accuracy, our research demonstrates that both can be achieved simultaneously with RGB input alone. The key idea of our approach is to enhance the ability for geometry estimation by combining easy-to-obtain monocular priors with learning-based dense SLAM, and then using 3D Gaussian splatting as our core map representation to efficiently model the scene. Upon loop closure, our method ensures on-the-fly global consistency through efficient pose graph bundle adjustment and instant map updates by explicitly deforming the 3D Gaussian units based on anchored keyframe updates. Furthermore, we introduce a grid-based scale alignment strategy to maintain improved scale consistency in prior depths for finer depth details. Through extensive experiments on Replica, ScanNet, Waymo Open, ETH3D SLAM and ScanNet++ datasets, we demonstrate significant improvements over existing Neural SLAM methods and even surpass RGB-D-based methods in both reconstruction and rendering quality.

We propose an approach based on convex relaxations for certifiably optimal robust multiview triangulation. To this end, we extend existing relaxation approaches to non-robust multiview triangulation by incorporating a least squares cost function. We propose two formulations, one based on epipolar constraints and one based on fractional reprojection constraints. The first is lower dimensional and remains tight under moderate noise and outlier levels, while the second is higher dimensional and therefore slower but remains tight even under extreme noise and outlier levels. We demonstrate through extensive experiments that the proposed approaches allow us to compute provably optimal re-constructions even under significant noise and a large percentage of outliers.

In this paper, we present a complete pipeline for 3D semantic mapping solely based on a stereo camera system. The pipeline comprises a direct sparse visual odometry frontend as well as a back-end for global optimization including GNSS integration, and semantic 3D point cloud labeling. We propose a simple but effective temporal voting scheme which improves the quality and consistency of the 3D point labels. Qualitative and quantitative evaluations of our pipeline are performed on the KITTI-360 dataset. The results show the effectiveness of our proposed voting scheme and the capability of our pipeline for efficient large-scale 3D semantic mapping. The large-scale mapping capabilities of our pipeline is furthermore demonstrated by presenting a very large-scale semantic map covering 8000 km of roads generated from data collected by a fleet of vehicles.

In this paper, we propose MonoRec, a semi-supervised monocular dense reconstruction architecture that predicts depth maps from a single moving camera in dynamic environments. MonoRec is based on a multi-view stereo setting which encodes the information of multiple consecutive images in a cost volume. To deal with dynamic objects in the scene, we introduce a MaskModule that predicts moving object masks by leveraging the photometric inconsistencies encoded in the cost volumes. Unlike other multi-view stereo methods, MonoRec is able to reconstruct both static and moving objects by leveraging the predicted masks. Furthermore, we present a novel multi-stage training scheme with a semi-supervised loss formulation that does not require LiDAR depth values. We carefully evaluate MonoRec on the KITTI dataset and show that it achieves state-of-theart performance compared to both multi-view and singleview methods. With the model trained on KITTI, we furthermore demonstrate that MonoRec is able to generalize well to both the Oxford RobotCar dataset and the more challenging TUM-Mono dataset recorded by a handheld camera. Code and related materials are available at https://vision.in.tum.de/research/monorec.

In this paper we propose a framework for inte-grating map-based relocalization into online direct visual odometry. To achieve map-based relocalization for direct methods, we integrate image features into Direct Sparse Odometry (DSO) and rely on feature matching to associate online visual odometry (VO) with a previously built map. The integration of the relocalization poses is threefold. Firstly, they are incorporated as pose priors in the direct image alignment of the front-end tracking. Secondly, they are tightly integrated into the back-end bundle adjustment. Thirdly, an online fusion module is further proposed to combine relative VO poses and global relocalization poses in a pose graph to estimate keyframe-wise smooth and globally accurate poses. We evaluate our method on two multi-weather datasets showing the benefits of integrating different handcrafted and learned features and demonstrating promising improvements on camera tracking accuracy.

In this paper we propose a novel method for depth estimation based on a single recording of a focused plenoptic camera. The presented algorithm is based on multiple stereo-observations within the multi-view micro images of the focused plenoptic camera. Here, pixel correspondences are found based on local intensity error minimization. Since our algorithm works directly on the micro images, no sub-aperture or epipolar plane images have to be synthesized. Due to the fact that we perform stereo matching based on local criteria we only estimate depth for pixels with sufficient gradient. Thus, we reduce the complexity of the problem, while neglecting uncertain stereo correspondences. Our algorithm incorporates multiple stereo-observations of the same point in a probabilistic depth map. We will show, that this (inverse) depth map can be modeled as a map of Gaussian distributed random variables. Thus, each depth pixel consists of an estimated depth and a corresponding variance, which gives a measure for the uncertainty of the estimation. This uncertainty information can be used in subsequent filtering methods.