Für eine korrekte Darstellung dieser Seite benötigen Sie einen XHTML-standardkonformen Browser, der die Darstellung von CSS-Dateien zulässt.

News

You are here



Funded Projects:

DFG Research Unit 5336 - Learning2Sense

DeToL – Deep Topology Learning - funded by the BMBF (Federal Ministry of Education and Research)

Video Segmentation from Multiple Representations using Lifted Multicuts - DFG Project KE 2264/1-1

ClimateVisions: automated image analysis of Reactions and Emotions from Images on Climate Change in Social Media

more details coming soon ...

Robustness and NAS 
 

Further Recent Projects on Robustness:

Trading off Image Quality for Robustness is not Necessary with Regularized Deterministic Autoencoders

Robust Models are less Over-Confident

Impact of realistic properties of the point spread function on classification tasks to reveal a possible distribution shift

Aliasing and adversarial robust generalization of CNNs

Intra-Source Style Augmentation for Improved Domain Generalization

FrequencyLowCut Pooling--Plug & Play against Catastrophic Overfitting

Estimating the Robustness of Classification Models by the Structure of the Learned Feature-Space

SpectralDefense: Detecting Adversarial Attacks on CNNs in the Fourier Domain

Is RobustBench/AutoAttack a suitable Benchmark for Adversarial Robustness?

Detecting AutoAttack Perturbations in the Frequency Domain

 

Image Generation and DeepFake Detection:

Internalized biases in fréchet inception distance

Shape your Space: A Gaussian Mixture Regularization Approach to Deterministic Autoencoders

Multi-Class Multi-Instance Count Conditioned Adversarial Image Generation

Beyond the Spectrum: Detecting Deepfakes via Re-Synthesis

Watch your up-convolution: Cnn based generative deep neural networks are failing to reproduce spectral distributions

Spectral Distribution aware Image Generation

Spectral Distribution Aware Image Generation | Papers With Code
 

Neural Architecture Search:

Learning Where To Look--Generative NAS is Surprisingly Efficient

DARTS for inverse problems: A study on stability

Smooth Variational Graph Embeddings for Efficient Neural Architecture Search

NAS-Bench-301 and the Case for Surrogate Benchmarks for Neural Architecture Search

Surrogate NAS benchmarks: Going beyond the limited search spaces of tabular NAS benchmarks

Neural Architecture Performance Prediction Using Graph Neural Networks

Figure 1 for Learning Where To Look -- Generative NAS is Surprisingly Efficient
 

Motion Segmentation and Tracking:

Higher-Order Multicuts for Geometric Model Fitting and Motion Segmentation

Uncertainty in Minimum Cost Multicuts for Image and Motion Segmentation

Motion Segmentation & Multiple Object Tracking by Correlation Co-Clustering

Motion trajectory segmentation via minimum cost multicuts

Object Segmentation Tracking from Generic Video Cues

A Two-Stage Minimum Cost Multicut Approach to Self-Supervised Multiple Person Tracking

Self-supervised Sparse to Dense Motion Segmentation

Motion Segmentation & Multiple Object Tracking by Correlation  Co-Clustering | Max Planck Institute for Intelligent Systems

Projects on Efficient Image Decompsition and Clustering:

MSM: Multi-stage Multicuts for Scalable Image Clustering

Learning embeddings for image clustering: an empirical study of triplet loss approaches

Learning to solve Minimum Cost Multicuts efficiently using Edge-Weighted Graph Convolutional Neural Networks

Optimizing Edge Detection for Image Segmentation with Multicut Penalties

Efficient Decomposition of Image and Mesh Graphs by Lifted Multicuts

A Benders Decomposition Approach to Correlation Clustering