Vision foundation models (VFMs) trained on large-scale image datasets provide high-quality features that have significantly advanced$2 D$visual recognition. However, their potential in 3D scene segmentation remains largely untapped, despite the common availability of$2 D$images alongside 3D point cloud datasets. While significant research has been dedicated to 2D-3D fusion, recent state-of-the-art 3D methods predominantly focus on 3D data, leaving the integration of VFMs into 3D models underexplored. In this work, we challenge this trend by introducing DITR, a generally applicable approach that extracts$2 D$foundation model features, projects them to 3D, and finally injects them into a 3D point cloud segmentation model. DITR achieves state-of-the-art results on both indoor and outdoor 3D semantic segmentation benchmarks. To enable the use of VFMs even when images are unavailable during inference, we additionally propose to pretrain 3D models by distilling 2D foundation models. By initializing the 3D backbone with knowledge distilled from 2D VFMs, we create a strong basis for downstream 3D segmentation tasks, ultimately boosting performance across various datasets.

Fisheye cameras offer robots the ability to capture human movements across a wider field of view (FOV) than standard pinhole cameras, making them particularly useful for applications in human-robot interaction and automotive contexts. However, accurately detecting human poses in fisheye images is challenging due to the curved distortions inherent to fisheye optics. While various methods for undistorting fisheye images have been proposed, their effectiveness and limitations for poses that cover a wide FOV has not been systematically evaluated in the context of absolute human pose estimation from monocular fisheye images. To address this gap, we evaluate the impact of pinhole, equidistant and double sphere camera models, as well as cylindrical projection methods, on 3D human pose estimation accuracy. We find that in close-up scenarios, pinhole projection is inadequate, and the optimal projection method varies with the FOV covered by the human pose. The usage of advanced fisheye models like the double sphere model significantly enhances 3D human pose estimation accuracy. We propose a heuristic for selecting the appropriate projection model based on the detection bounding box to enhance prediction quality. Additionally, we introduce and evaluate on our novel FISHnCHIPS dataset, which features 3D human skeleton annotations in fisheye images, including images from unconventional angles, such as extreme close-ups, ground-mounted cameras, and wide-FOV poses, available at: https://www.vision.rwth-aachen.de/fishnchips.

Robot manipulation relying on learned object-centric descriptors became popular in recent years. Visual descriptors can easily describe manipulation task objectives, they can be learned efficiently using self-supervision, and they can encode actuated and even non-rigid objects. However, learning robust, view-invariant keypoints in a self-supervised approach requires a meticulous data collection approach involving precise calibration and expert supervision. In this paper we introduce Cycle-Correspondence Loss (CCL) for view-invariant dense descriptor learning, which adopts the concept of cycle-consistency, enabling a simple data collection pipeline and training on unpaired RGB camera views. The key idea is to autonomously detect valid pixel correspondences by attempting to use a prediction over a new image to predict the original pixel in the original image, while scaling error terms based on the estimated confidence. Our evaluation shows that we outperform other self-supervised RGB-only methods, and approach performance of supervised methods, both with respect to keypoint tracking as well as for a robot grasping downstream task.

We propose a framework for robust and efficient training of Dense Object Nets (DON) [1] with a focus on industrial multi-object robot manipulation scenarios. DON is a popular approach to obtain dense, view-invariant object descriptors, which can be used for a multitude of downstream tasks in robot manipulation, such as, pose estimation, state representation for control, etc. However, the original work [1] focused training on singulated objects, with limited results on instance-specific, multi-object applications. Additionally, a complex data collection pipeline, including 3D reconstruction and mask annotation of each object, is required for training. In this paper, we further improve the efficacy of DON with a simplified data collection and training regime, that consistently yields higher precision and enables robust tracking of keypoints with less data requirements. In particular, we focus on training with multi-object data instead of singulated objects, combined with a well-chosen augmentation scheme. We additionally propose an alternative loss formulation to the original pixel wise formulation that offers better results and is less sensitive to hyperparameters. Finally, we demonstrate the robustness and accuracy of our proposed framework on a real-world robotic grasping task.

We present the D-eDVS- a combined event-based 3D sensor - and a novel event-based full-3D simultaneous localization and mapping algorithm which works exclusively with the sparse stream of visual data provided by the D-eDVS. The D-eDVS is a combination of the established PrimeSense RGB-D sensor and a biologically inspired embedded dynamic vision sensor. Dynamic vision sensors only react to dynamic contrast changes and output data in form of a sparse stream of events which represent individual pixel locations. We demonstrate how an event-based dynamic vision sensor can be fused with a classic frame-based RGB-D sensor to produce a sparse stream of depth-augmented 3D points. The advantages of a sparse, event-based stream are a much smaller amount of generated data, thus more efficient resource usage, and a continuous representation of motion allowing lag-free tracking. Our event-based SLAM algorithm is highly efficient and runs 20 times faster than realtime, provides localization updates at several hundred Hertz, and produces excellent results. We compare our method against ground truth from an external tracking system and two state-of-the-art algorithms on a new dataset which we release in combination with this paper.