TUM-VIE: The TUM Stereo Visual-Inertial Event Dataset

Klenk, Simon; Chui, Jason; Demmel, Nikolaus; Cremers, Daniel

doi:10.1109/iros51168.2021.9636728

Cited by 35 publications

(25 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Publicly available event-based Visual Odometry (VO) datasets are limited and/or contain low quality sensory data [50,59], either because of noisy events recorded with pioneer neuromorphic devices (such as the DAVIS240B [60] or DAVIS346B) or due to low quality frames (i.e., no gamma correction, small fill factor). Recent publicly available datasets, such as [61,62], are equipped with newer sensors, however they do not share the same optical axis, having frames and events in two image planes, standard and event-based camera separately. There is a clear need for up-to-date datasets in the event-based VO community with state-of-the-art sensors.…”

Section: Beamsplitter and Additional Experimentsmentioning

confidence: 99%

Event-aided Direct Sparse Odometry

Hidalgo-Carrió¹,

Gallego²,

Scaramuzza³

2022

Preprint

View full text Add to dashboard Cite

show abstract

Section: Beamsplitter and Additional Experimentsmentioning

confidence: 99%

Event-aided Direct Sparse Odometry

Hidalgo-Carrió¹,

Gallego²,

Scaramuzza³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Due to technical reasons, GPS is not included in the data, and ground truth is approximated by three different LiDAR SLAM algorithms. Finally, the TUM-VIE [18] dataset is captured by a pair of Prophesee Gen4 CD event cameras (1280×720, events only), a stereo camera, and a six-axis IMU. The sequences are recorded in differently scaled environments and under different motion conditions (e.g.…”

Section: Related Workmentioning

confidence: 99%

“…The event stereo cameras have VGA resolution (640 × 480) with a horizontal baseline of about 17cm. Since the MoCap system emits 850nm infrared strobes to locate the passive spherical markers, we have adopted the common practice of putting an infrared filter (PHTODE IR690, cutoff frequency of 400-690nm) in front of the lenses [18], [21], thus blocking most of the flashing and reducing the background noise in the event stream. The reason for choosing a pair of event cameras with VGA (640 × 480) rather than HD resolution (1280 × 720) is due to the fact that we have observed a smearing effect on top of the surface of active events.…”

Section: A Sensor Setupmentioning

confidence: 99%

VECtor: A Versatile Event-Centric Benchmark for Multi-Sensor SLAM

Gao,

Liang,

Yang

et al. 2022

Preprint

View full text Add to dashboard Cite

Event cameras have recently gained in popularity as they hold strong potential to complement regular cameras in situations of high dynamics or challenging illumination. An important problem that may benefit from the addition of an event camera is given by Simultaneous Localization And Mapping (SLAM). However, in order to ensure progress on event-inclusive multi-sensor SLAM, novel benchmark sequences are needed. Our contribution is the first complete set of benchmark datasets captured with a multi-sensor setup containing an event-based stereo camera, a regular stereo camera, multiple depth sensors, and an inertial measurement unit. The setup is fully hardwaresynchronized and underwent accurate extrinsic calibration. All sequences come with ground truth data captured by highly accurate external reference devices such as a motion capture system. Individual sequences include both small and large-scale environments, and cover the specific challenges targeted by dynamic vision sensors.

show abstract

“…Datasets. We evaluate our stereo methods on sequences from five publicly available datasets [27], [38], [54]- [56] and a simulator. Sequences from [38], [55] were acquired with a hand-held stereo or trinocular event camera in indoor environments.…”

Section: A Datasets and Evaluation Metricsmentioning

confidence: 99%

“…We present depth estimation results using Alg. 1 on the TUM-VIE dataset [56], the first public visual-inertial dataset with 1 Megapixel stereo event cameras (Prophesee Gen4 [5]). To the best of our knowledge, our work is the first to provide results on this new event-based dataset (the original paper presented the data but did not evaluate it on any event-based algorithm).…”

Section: F Experiments On Tum-vie Datasetmentioning

confidence: 99%

Multi-Event-Camera Depth Estimation and Outlier Rejection by Refocused Events Fusion

Ghosh¹,

Gallego²

2022

Preprint

View full text Add to dashboard Cite

Event cameras are bio-inspired sensors that offer advantages over traditional cameras. They work asynchronously, sampling the scene with microsecond resolution and producing a stream of brightness changes. This unconventional output has sparked novel computer vision methods to unlock the camera's potential. We tackle the problem of event-based stereo 3D reconstruction for SLAM. Most event-based stereo methods try to exploit the camera's high temporal resolution and event simultaneity across cameras to establish matches and estimate depth. By contrast, we investigate how to estimate depth without explicit data association by fusing Disparity Space Images (DSIs) originated in efficient monocular methods. We develop fusion theory and apply it to design multi-camera 3D reconstruction algorithms that produce state-of-the-art results, as we confirm by comparing against four baseline methods and testing on a variety of available datasets.

show abstract

TUM-VIE: The TUM Stereo Visual-Inertial Event Dataset

Cited by 35 publications

References 12 publications

Event-aided Direct Sparse Odometry

Event-aided Direct Sparse Odometry

VECtor: A Versatile Event-Centric Benchmark for Multi-Sensor SLAM

Multi-Event-Camera Depth Estimation and Outlier Rejection by Refocused Events Fusion

Contact Info

Product

Resources

About