25 skills found
naver / Kapturekapture is a file format as well as a set of tools for manipulating datasets, and in particular Visual Localization and Structure from Motion data.
cvdfoundation / Ava DatasetThe AVA dataset densely annotates 80 atomic visual actions in 351k movie clips with actions localized in space and time, resulting in 1.65M action labels with multiple labels per human occurring frequently.
IntelliSensing / UAV VisLocUAV-VisLoc: A Large-scale Dataset for UAV Visual Localization
cvg / Visloc Iccv2021ETH-Microsoft dataset for the ICCV 2021 visual localization challenge
ardasnck / Learning To Localize Sound SourceCodebase and Dataset for the paper: Learning to Localize Sound Source in Visual Scenes
facebookresearch / 3D Vision And TouchWhen told to understand the shape of a new object, the most instinctual approach is to pick it up and inspect it with your hand and eyes in tandem. Here, touch provides high fidelity localized information while vision provides complementary global context. However, in 3D shape reconstruction, the complementary fusion of visual and haptic modalities remains largely unexplored. In this paper, we study this problem and present an effective chart-based approach to fusing vision and touch, which leverages advances in graph convolutional networks. To do so, we introduce a dataset of simulated touch and vision signals from the interaction between a robotic hand and a large array of 3D objects. Our results show that (1) leveraging both vision and touch signals consistently improves single-modality baselines, especially when the object is occluded by the hand touching it; (2) our approach outperforms alternative modality fusion methods and strongly benefits from the proposed chart-based structure; (3) reconstruction quality boosts with the number of grasps provided; and (4) the touch information not only enhances the reconstruction at the touch site but also extrapolates to its local neighborhood.
hmf21 / AerialVLAerialVL: a Visual Geo-Localization Dataset Designed for Aerial-based Vehicles
neis-lab / MmcowsMmCows: A Multimodal Dataset for Dairy Cattle Monitoring
remaro-network / SubPipe DatasetA Submarine Pipeline Inspection Dataset for Segmentation and Visual-inertial Localization
alexmelekhin / HPointLocHPointLoc: open dataset and framework for indoor visual localization based on synthetic RGB-D images
HuajianUP / 360Loc[CVPR' 24] Toolkit for 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries Resources
TIERS / Uwb Vio Lidar FusionA dataset and basic ROS nodes for localization in GNSS-Denied environments with 3D lidar, visual-inertial odometry (T265 camera) and ultra-wideband modules (DWM1001 with Decawave's and custom firmware)
jrcuaranv / Terrasentia DatasetThis dataset is intended for the evaluation of visual-based localization and mapping systems in agriculture.
leaderj1001 / Action LocalizationAction-Localization, Atomic Visual Actions (AVA) Dataset
scutzetao / DLfeature PlaceRecog Icra2017Dataset and official Caffe Implementation for learning a condition-robust feature representation for long-term visual localization
kimjh069 / CLocDatasets for long-term visual localization with sequential images in large-scale spaces
pit30m / Pit30mThe Python SDK for the Pit30M large scale visual localization dataset.
syywh / YQ5Long-term visual localization datasets in campus environment
cds-mipt / HPointLocOpen dataset and framework for visual place recognition and localization
maazmb / LEP Hybrid Visual OdometryWe propose a novel real time monocular Hybrid visual odometry formulation which combines the high precision of indirect approaches with the fast performance of direct methods. The system initializes inverse depth estimates represented as a Gaussian probability distribution for features (lines, edges and points) extracted in each keyframe which we continuously propagate and update with new measurements in the following frames. The key idea is to incorporate the depth filter distributions into the initial pose tracking via sparse image alignment and also the pose refinement via map localization. We also propose a comprehensive initialization method of these depth filters and classify the map points into different categories based on the uncertainty of these depth estimates which as a result greatly improves the tracking performance. The experimental evaluation on benchmark datasets shows that the proposed approach is significantly faster than the state-of-the-art algorithms while achieving comparable accuracy. We make our implementation publically open source at github to provide as a valuable reference for the SLAM community.
