no code implementations • 8 Apr 2024 • Ernst Seidel, Pejman Mowlaee, Tim Fingscheidt
In recent years, the introduction of neural networks (NNs) into the field of speech enhancement has brought significant improvements.
1 code implementation • 4 Dec 2023 • Joshua Niemeijer, Manuel Schwonberg, Jan-Aike Termöhlen, Nico M. Schmidt, Tim Fingscheidt
In a second step, we train a generalizing model by adapting towards this pseudo-target domain.
no code implementations • 5 Sep 2023 • Ziyi Xu, Marvin Sach, Jan Pirklbauer, Tim Fingscheidt
It provides a reference-free perceptual loss for employing real data during DNS training, maximizing the PESQ scores.
1 code implementation • 25 Aug 2023 • Jan-Aike Termöhlen, Timo Bartels, Tim Fingscheidt
We present a new augmentation-driven approach to domain generalization for semantic segmentation using a re-parameterized vision transformer (ReVT) with weight averaging of multiple models after training.
no code implementations • 28 Jul 2023 • Ernst Seidel, Pejman Mowlaee, Tim Fingscheidt
The topic of deep acoustic echo control (DAEC) has seen many approaches with various model topologies in recent years.
1 code implementation • Open Journal on ITS 2023 • Christopher Plachetka, Benjamin Sertolli, Jenny Fricke, Marvin Klingner, Tim Fingscheidt
In this work we present a novel deep learning-based approach to detect and specify map deviations in erroneous or outdated high-definition (HD) maps using both sensor and map data as input to a deep neural network (DNN).
no code implementations • 5 Jun 2023 • Marvin Sach, Jan Franzen, Bruno Defraene, Kristoff Fluyt, Maximilian Strake, Wouter Tirry, Tim Fingscheidt
By applying a number of topological changes at once, we propose both an efficient FCRN (FCRN15), and a new family of efficient convolutional recurrent neural networks (EffCRN23, EffCRN23lite).
no code implementations • 24 Apr 2023 • Manuel Schwonberg, Joshua Niemeijer, Jan-Aike Termöhlen, Jörg P. Schäfer, Nico M. Schmidt, Hanno Gottschalk, Tim Fingscheidt
DNNs play a significant role in environment perception for the challenging application of automated driving and are employed for tasks such as detection, semantic segmentation, and sensor fusion.
1 code implementation • 18 Apr 2023 • Ziyi Xu, Ziyue Zhao, Tim Fingscheidt
We illustrate the potential of this model by predicting the PESQ scores of wideband-coded speech obtained from AMR-WB or EVS codecs operating at different bitrates in noisy, tandeming, and error-prone transmission conditions.
1 code implementation • 20 Sep 2022 • Timo Lohrenz, Björn Möller, Zhengyang Li, Tim Fingscheidt
The powerful modeling capabilities of all-attention-based transformer architectures often cause overfitting and - for natural language processing tasks - lead to an implicitly learned internal language model in the autoregressive transformer decoder complicating the integration of external language models.
Ranked #4 on Lipreading on LRS3-TED (using extra training data)
1 code implementation • 1 Jun 2022 • Jasmin Breitenstein, Tim Fingscheidt
In this paper, we consider the task of amodal semantic segmentation and propose a generic way to generate datasets to train amodal semantic segmentation methods.
no code implementations • 1 Jun 2022 • Marvin Klingner, Konstantin Müller, Mona Mirzaie, Jasmin Breitenstein, Jan-Aike Termöhlen, Tim Fingscheidt
The emergence of data-driven machine learning (ML) has facilitated significant progress in many complicated tasks such as highly-automated driving.
no code implementations • 9 May 2022 • Ernst Seidel, Rasmus Kongsgaard Olsson, Karim Haddad, Zhengyang Li, Pejman Mowlaee, Tim Fingscheidt
Although today's speech communication systems support various bandwidths from narrowband to super-wideband and beyond, state-of-the art DNN methods for acoustic echo cancellation (AEC) are lacking modularity and bandwidth scalability.
no code implementations • 4 May 2022 • Ziyi Xu, Maximilian Strake, Tim Fingscheidt
Detailed analyses show that the DNS trained with the MF-intrusive PESQNet outperforms the Interspeech 2021 DNS Challenge baseline and the same DNS trained with an MSE loss by 0. 23 and 0. 12 PESQ points, respectively.
1 code implementation • 2 Mar 2022 • Marvin Klingner, Mouadh Ayache, Tim Fingscheidt
In this work, we further expand a source-free UDA approach to a continual and therefore online-capable UDA on a single-image basis for semantic segmentation.
1 code implementation • 2 Mar 2022 • Marvin Klingner, Varun Ravi Kumar, Senthil Yogamani, Andreas Bär, Tim Fingscheidt
In this paper, we (i) propose a novel adversarial perturbation detection scheme based on multi-task perception of complex vision tasks (i. e., depth estimation and semantic segmentation).
no code implementations • 17 Jan 2022 • Jonas Löhdefink, Tim Fingscheidt
The proposed methodology allows to achieve an mIoU improvement on the Cityscapes validation set of 5. 7% absolute over the same CycleGAN without noise injection, and still an absolute 4. 9% over the ERFNet non-cyclic baseline.
no code implementations • 8 Jan 2022 • Bile Peng, Jan-Aike Termöhlen, Cong Sun, Danping He, Ke Guan, Tim Fingscheidt, Eduard A. Jorswieck
The rectangular shape of the RIS and the spatial correlation of channels with adjacent RIS antennas due to the short distance between them encourage us to apply it for the RIS configuration.
no code implementations • 6 Nov 2021 • Ziyi Xu, Maximilian Strake, Tim Fingscheidt
Perceptual evaluation of speech quality (PESQ) is a widely used metric for evaluating speech quality.
no code implementations • 20 Sep 2021 • Daniel Bogdoll, Jasmin Breitenstein, Florian Heidecker, Maarten Bieshaar, Bernhard Sick, Tim Fingscheidt, J. Marius Zöllner
Scaling the distribution of automated vehicles requires handling various unexpected and possibly dangerous situations, termed corner cases (CC).
no code implementations • 6 Aug 2021 • Jan Franzen, Tim Fingscheidt
Deep neural network (DNN)-based approaches to acoustic echo cancellation (AEC) and hybrid speech enhancement systems have gained increasing attention recently, introducing significant performance improvements to this research field.
1 code implementation • 2 Jul 2021 • Timo Lohrenz, Patrick Schwarz, Zhengyang Li, Tim Fingscheidt
Recently, attention-based encoder-decoder (AED) models have shown high performance for end-to-end automatic speech recognition (ASR) across several tasks.
Ranked #7 on Speech Recognition on WSJ eval92
Automatic Speech Recognition Automatic Speech Recognition (ASR) +2
no code implementations • 29 Apr 2021 • Sebastian Houben, Stephanie Abrecht, Maram Akila, Andreas Bär, Felix Brockherde, Patrick Feifel, Tim Fingscheidt, Sujan Sai Gannamaneni, Seyed Eghbal Ghobadi, Ahmed Hammam, Anselm Haselhoff, Felix Hauser, Christian Heinzemann, Marco Hoffmann, Nikhil Kapoor, Falk Kappel, Marvin Klingner, Jan Kronenberger, Fabian Küppers, Jonas Löhdefink, Michael Mlynarski, Michael Mock, Firas Mualla, Svetlana Pavlitskaya, Maximilian Poretschkin, Alexander Pohl, Varun Ravi-Kumar, Julia Rosenzweig, Matthias Rottmann, Stefan Rüping, Timo Sämann, Jan David Schneider, Elena Schulz, Gesina Schwalbe, Joachim Sicking, Toshika Srivastava, Serin Varghese, Michael Weber, Sebastian Wirkert, Tim Wirtz, Matthias Woehrle
Our paper addresses both machine learning experts and safety engineers: The former ones might profit from the broad range of machine learning topics covered and discussions on limitations of recent methods.
no code implementations • 12 Apr 2021 • Marvin Klingner, Andreas Bär, Marcel Mross, Tim Fingscheidt
In this work we address the task of observing the performance of a semantic segmentation deep neural network (DNN) during online operation, i. e., during inference, which is of high importance in safety-critical applications such as autonomous driving.
no code implementations • 9 Apr 2021 • Varun Ravi Kumar, Marvin Klingner, Senthil Yogamani, Markus Bach, Stefan Milz, Tim Fingscheidt, Patrick Mäder
We evaluate our approach on the Fisheye WoodScape surround-view dataset, significantly improving over previous approaches.
no code implementations • 31 Mar 2021 • Timo Lohrenz, Zhengyang Li, Tim Fingscheidt
Stream fusion, also known as system combination, is a common technique in automatic speech recognition for traditional hybrid hidden Markov model approaches, yet mostly unexplored for modern deep neural network end-to-end model architectures.
Automatic Speech Recognition Automatic Speech Recognition (ASR) +3
no code implementations • 31 Mar 2021 • Ziyi Xu, Maximilian Strake, Tim Fingscheidt
During the training process, most of the speech enhancement neural networks are trained in a fully supervised way with losses requiring noisy speech to be synthesized by clean speech and additive noise.
no code implementations • 31 Mar 2021 • Ernst Seidel, Jan Franzen, Maximilian Strake, Tim Fingscheidt
The proposed models achieved remarkable performance for the separate tasks of AEC and residual echo suppression (RES).
1 code implementation • 16 Mar 2021 • Jan Franzen, Ernst Seidel, Tim Fingscheidt
Acoustic echo cancellation (AEC) algorithms have a long-term steady role in signal processing, with approaches improving the performance of applications such as automotive hands-free systems, smart home and loudspeaker devices, or web conference systems.
no code implementations • 5 Mar 2021 • Florian Heidecker, Jasmin Breitenstein, Kevin Rösch, Jonas Löhdefink, Maarten Bieshaar, Christoph Stiller, Tim Fingscheidt, Bernhard Sick
Systems and functions that rely on machine learning (ML) are the basis of highly automated driving.
no code implementations • 11 Feb 2021 • Jasmin Breitenstein, Jan-Aike Termöhlen, Daniel Lipinski, Tim Fingscheidt
Hence, their detection is highly safety-critical, and detection methods can be applied to vast amounts of collected data to select suitable training data.
no code implementations • 11 Jan 2021 • Andreas Bär, Jonas Löhdefink, Nikhil Kapoor, Serin J. Varghese, Fabian Hüger, Peter Schlicht, Tim Fingscheidt
Although CNNs obtain state-of-the-art performance on clean images, almost imperceptible changes to the input, referred to as adversarial perturbations, may lead to fatal deception.
no code implementations • 2 Dec 2020 • Nikhil Kapoor, Andreas Bär, Serin Varghese, Jan David Schneider, Fabian Hüger, Peter Schlicht, Tim Fingscheidt
Despite recent advancements, deep neural networks are not robust against adversarial perturbations.
no code implementations • 2 Dec 2020 • Nikhil Kapoor, Chun Yuan, Jonas Löhdefink, Roland Zimmermann, Serin Varghese, Fabian Hüger, Nico Schmidt, Peter Schlicht, Tim Fingscheidt
Deep neural networks are often not robust to semantically-irrelevant changes in the input.
2 code implementations • 17 Nov 2020 • Marvin Klingner, Jan-Aike Termöhlen, Jacob Ritterbach, Tim Fingscheidt
In this paper we present a solution to the task of "unsupervised domain adaptation (UDA) of a given pre-trained semantic segmentation model without relying on any source domain representations".
no code implementations • 28 Oct 2020 • Atiye Sadat Hashemi, Andreas Bär, Saeed Mozaffari, Tim Fingscheidt
Using our generated non-targeted UAPs, we obtain an average fooling rate of 93. 36% on the source models (state of the art: 82. 16%).
no code implementations • 10 Aug 2020 • Varun Ravi Kumar, Marvin Klingner, Senthil Yogamani, Stefan Milz, Tim Fingscheidt, Patrick Maeder
This paper introduces a novel multi-task learning strategy to improve self-supervised monocular distance estimation on fisheye and pinhole camera images.
1 code implementation • 16 Jul 2020 • Antonia Breuer, Jan-Aike Termöhlen, Silviu Homoceanu, Tim Fingscheidt
Analyzing and predicting the traffic scene around the ego vehicle has been one of the key challenges in autonomous driving.
1 code implementation • ECCV 2020 • Marvin Klingner, Jan-Aike Termöhlen, Jonas Mikolajczyk, Tim Fingscheidt
Self-supervised monocular depth estimation presents a powerful method to obtain 3D scene information from single camera images, which is trainable on arbitrary image sequences without requiring depth labels, e. g., from a LiDAR sensor.
no code implementations • 15 Jun 2020 • Jonas Löhdefink, Justin Fehrling, Marvin Klingner, Fabian Hüger, Peter Schlicht, Nico M. Schmidt, Tim Fingscheidt
Autonomous driving requires self awareness of its perception functions.
1 code implementation • 12 May 2020 • Marvin Klingner, Andreas Bär, Philipp Donn, Tim Fingscheidt
While neural networks trained for semantic segmentation are essential for perception in autonomous driving, most current algorithms assume a fixed number of classes, presenting a major limitation when developing new autonomous driving systems with the need of additional classes.
no code implementations • 23 Apr 2020 • Marvin Klingner, Andreas Bär, Tim Fingscheidt
We show the effectiveness of our method on the Cityscapes dataset, where our multi-task training approach consistently outperforms the single-task semantic segmentation baseline in terms of both robustness vs. noise and in terms of adversarial attacks, without the need for depth labels in training.
1 code implementation • 23 May 2019 • Ziyue Zhao, Samy Elshamy, Tim Fingscheidt
Single-channel speech enhancement with deep neural networks (DNNs) has shown promising performance and is thus intensively being studied.
no code implementations • 25 Feb 2019 • Jan-Aike Bolte, Andreas Bär, Daniel Lipinski, Tim Fingscheidt
The progress in autonomous driving is also due to the increased availability of vast amounts of training data for the underlying machine learning approaches.
no code implementations • 12 Feb 2019 • Jonas Löhdefink, Andreas Bär, Nico M. Schmidt, Fabian Hüger, Peter Schlicht, Tim Fingscheidt
The high amount of sensors required for autonomous driving poses enormous challenges on the capacity of automotive bus systems.
1 code implementation • 25 Jun 2018 • Ziyue Zhao, Huijun Liu, Tim Fingscheidt
Enhancing coded speech suffering from far-end acoustic background noise, quantization noise, and potentially transmission errors, is a challenging task.