Our paper "Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference" has been accepted to #ECCV2020. We reparameterize the convs to eliminate task interference and allow for the incremental learning of new tasks. Paper: arxiv.org/abs/2007.12540

Join us today for the poster session of our work "Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference" at #ECCV2020 ! Session 1: Wednesday 26 August: 06:00 - 08:00 (UTC+1) Session 2: Wednesday 26 August: 14:00 -16:00 (UTC+1)

Check out our #BMVC2020 paper "Automated Search for Resource-Efficient Branched Multi-Task Networks". We propose an approach to automatically define branched multi-task networks, while using a resource-aware loss to control the model size Paper: arxiv.org/abs/2008.10292

The codebase from our #ECCV2020 paper titled "Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference" is now publicly available: github.com/menelaoskanaki…

We have updated our survey on multi-task learning for dense prediction tasks. The paper features an extensive literature review, intuitive comparisons, thorough experiments, etc. arxiv.org/abs/2004.13379 Code can be found here 🥳: github.com/SimonVandenhen…

Code packaging frenzy continues! Check out my latest python package, democratizing orthogonal transformations in #PyTorch. Goomba has nothing to do with the Householder transformation; your attention is all I need. github.com/toshas/torch-h… obukhov.ai/torch-househol…

Spectral Tensor Train Parameterization of Deep Learning Layers deepai.org/publication/sp… by Anton Obukhov et al. including Stamatios Georgoulis #NeuralNetwork #Tensor

1/ Our paper “Spectral Tensor Train Parameterization of Deep Learning Layers” about end-to-end neural network compression and stability of training in the GAN setting is live at #AISTATS2021 this week! PDF: arxiv.org/abs/2103.04217 Project page: obukhov.ai/sttp

Learning to Relate Depth and Semantics for Unsupervised Domain Adaptation deepai.org/publication/le… by Suman Saha et al. including Menelaos Kanakis, Stamatios Georgoulis #Estimator #Statistics

We study how biases in the dataset affect contrastive pretraining and explore additional invariances What if we use non-curated data (COCO, OpenImages) vs ImageNet? Do we need priors to learn dense representations? Paper: arxiv.org/abs/2106.05967 Code: github.com/wvangansbeke/R… 👇

We release our newest work on event cameras: "Time Lens". We use events to upsample low-framerate RGB HD video by over 50 times with only 1/40th of the memory footprint! #CVPR2021 Paper, code, datasets: rpg.ifi.uzh.ch/timelens Daniel Gehrig Mathias Gehrig

I will present our latest #AISTATS work about sparse and stable CNN parameterizations (pinned) in the sparsity workshop today in 4 hrs #SNN21

The source code of our #ICML 2020 paper "T-Basis: a Compact Representation for Neural Networks" is now on GitHub: github.com/toshas/tbasis

So honored that our #CVPR2021 on video frame interpolation with event cameras is on #twominutepapers! Daniel Gehrig Stepan Tulyakov

Our #ICCV2021 paper on Adaptive Task-Relational Context, leveraging neural architecture search and attention mechanisms in the multi-task learning setting, is now paired with source code! Menelaos Kanakis Stamatios Georgoulis paper: arxiv.org/abs/2104.13874 code: github.com/brdav/atrc

Excited to be hosting the DeepMTL workshop on multi-task learning at ICCV2021 tomorrow. We have an excellent group of speakers ready for you: Amir Zamir I. Kokkinos Judy Hoffman Raquel Urtasun R. Caruana A. Rabinovich sites.google.com/view/deepmtlwo…

Multi-Bracket High Dynamic Range Imaging with Event Cameras deepai.org/publication/mu… by Nico Messikommer et al. including Stamatios Georgoulis #ComputerVision #PatternRecognition

Motivated MSc/BSc students & prospective PhD candidates can always reach out to me -- plz see the contact instructions on my website. Plz help me spread the word: 🆘 🆘 🆘 We are actively hiring 1 PhD candidate together with Theo Gevers! Separate tweet coming soon 📢 📢 📢 (8/8)

1/n Our last work on event-based image deblurring has won the best paper award in the #MIPI workshop at #ECCV2022.

📢 Our new paper GaVS – 3D-Grounded Video Stabilization is out! Key idea: feed-forward Dynamic Gaussian Splatting + test-time optimization Robust, consistent, and cropping-free 📹 🎥 Project: sinoyou.github.io/gavs Zinuo You Stamatios Georgoulis Siyu Tang @VLG-ETHZ Dengxin Dai #SIGGRAPH25 #3DGS