1 code implementation • 23 Feb 2024 • Nikola Zubić, Mathias Gehrig, Davide Scaramuzza
We address this challenge by introducing state-space models (SSMs) with learnable timescale parameters to event-based vision.
1 code implementation • 29 Nov 2023 • Ziyi Wu, Mathias Gehrig, Qing Lyu, Xudong Liu, Igor Gilitschenski
On 1Mpx, RVT-S with 10% labels even surpasses its fully-supervised counterpart using 100% labels.
1 code implementation • 12 Jun 2023 • Yifei Liu, Mathias Gehrig, Nico Messikommer, Marco Cannici, Davide Scaramuzza
In relation to the dense counterpart that utilizes all tokens, our method realizes an increase in inference speed, achieving up to 34% faster performance for the entire network and 46% for the backbone.
1 code implementation • ICCV 2023 • Nikola Zubić, Daniel Gehrig, Mathias Gehrig, Davide Scaramuzza
However, selecting the appropriate representation for the task traditionally requires training a neural network for each representation and selecting the best one based on the validation score, which is very time-consuming.
Ranked #1 on Object Detection on GEN1 Detection
no code implementations • 14 Apr 2023 • Yannick Schnider, Stanislaw Wozniak, Mathias Gehrig, Jules Lecomte, Axel von Arnim, Luca Benini, Davide Scaramuzza, Angeliki Pantazi
Optical flow provides information on relative motion that is an important component in many computer vision pipelines.
1 code implementation • 24 Mar 2023 • Asude Aydin, Mathias Gehrig, Daniel Gehrig, Davide Scaramuzza
Our hybrid ANN-SNN model thus combines the best of both worlds: It does not suffer from long state transients and state decay thanks to the ANN, and can generate predictions with high temporal resolution, low latency, and low power thanks to the SNN.
1 code implementation • CVPR 2023 • Mathias Gehrig, Davide Scaramuzza
By revisiting the high-level design of recurrent vision backbones, we reduce inference time by a factor of 6 while retaining similar performance.
1 code implementation • CVPR 2023 • Nico Messikommer, Carter Fang, Mathias Gehrig, Davide Scaramuzza
Because of their high temporal resolution, increased resilience to motion blur, and very sparse output, event cameras have been shown to be ideal for low-latency and low-bandwidth feature tracking, even in challenging scenarios.
1 code implementation • 25 Mar 2022 • Mathias Gehrig, Manasi Muglikar, Davide Scaramuzza
To the best of our knowledge, our model is the first method that can regress dense pixel trajectories from event data.
1 code implementation • 6 Sep 2021 • Nico Messikommer, Daniel Gehrig, Mathias Gehrig, Davide Scaramuzza
However, event-based vision has been held back by the shortage of labeled datasets due to the novelty of event cameras.
1 code implementation • 24 Aug 2021 • Mathias Gehrig, Mario Millhäusler, Daniel Gehrig, Davide Scaramuzza
Modern frame-based optical flow methods heavily rely on matching costs computed from feature correlation.
1 code implementation • CVPR 2021 • Stepan Tulyakov, Daniel Gehrig, Stamatios Georgoulis, Julius Erbach, Mathias Gehrig, Yuanyou Li, Davide Scaramuzza
However, while these approaches can capture non-linear motions they suffer from ghosting and perform poorly in low-texture regions with few events.
1 code implementation • 14 Jun 2021 • Stepan Tulyakov, Daniel Gehrig, Stamatios Georgoulis, Julius Erbach, Mathias Gehrig, Yuanyou Li, Davide Scaramuzza
State-of-the-art frame interpolation methods generate intermediate frames by inferring object motions in the image from consecutive key-frames.
1 code implementation • 26 May 2021 • Manasi Muglikar, Mathias Gehrig, Daniel Gehrig, Davide Scaramuzza
We propose a generic event camera calibration framework using image reconstruction.
1 code implementation • 10 Mar 2021 • Mathias Gehrig, Willem Aarents, Daniel Gehrig, Davide Scaramuzza
To address these challenges, we propose, DSEC, a new dataset that contains such demanding illumination conditions and provides a rich set of sensory data.
1 code implementation • 18 Feb 2021 • Daniel Gehrig, Michelle Rüegg, Mathias Gehrig, Javier Hidalgo Carrio, Davide Scaramuzza
However, events only measure the varying component of the visual signal, which limits their ability to encode scene context.
no code implementations • 26 May 2020 • Philipp Foehn, Dario Brescianini, Elia Kaufmann, Titus Cieslewski, Mathias Gehrig, Manasi Muglikar, Davide Scaramuzza
This paper presents a novel system for autonomous, vision-based drone racing combining learned data abstraction, nonlinear filtering, and time-optimal trajectory planning.
1 code implementation • 5 Mar 2020 • Mathias Gehrig, Sumit Bam Shrestha, Daniel Mouritzen, Davide Scaramuzza
Due to their spike-based computational model, SNNs can process output from event-based, asynchronous sensors without any pre-processing at extremely lower power unlike standard artificial neural networks.
1 code implementation • CVPR 2020 • Daniel Gehrig, Mathias Gehrig, Javier Hidalgo-Carrió, Davide Scaramuzza
Event cameras are novel sensors that output brightness changes in the form of a stream of asynchronous "events" instead of intensity frames.
no code implementations • 11 Nov 2019 • Rika Sugimoto Dimitrova, Mathias Gehrig, Dario Brescianini, Davide Scaramuzza
In particular, this paper addresses the problem of one-dimensional attitude tracking using a dualcopter platform equipped with an event camera.
Robotics Systems and Control Systems and Control
1 code implementation • CVPR 2019 • Guillermo Gallego, Mathias Gehrig, Davide Scaramuzza
The proposed loss functions allow bringing mature computer vision tools to the realm of event cameras.
no code implementations • 11 Oct 2016 • Mathias Gehrig, Elena Stumm, Timo Hinzmann, Roland Siegwart
We propose a novel scoring concept for visual place recognition based on nearest neighbor descriptor voting and demonstrate how the algorithm naturally emerges from the problem formulation.