Lightnews — Scholar-powered news

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 24

Our #ICCV2025 paper will be presented as an Highlight ✨

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

Is Tracking really more challenging in First Person Egocentric Vision?

Our new #ICCV2025 paper follows up our IJCV 2023 study (bit.ly/4nVRJw9).

We further investigate the causes of performance drops in object and segmentation tracking under egocentric FPV settings.

🧵 (1/7)

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

Joint work with @zairamanigrasso.bsky.social and Christian Micheloni

Funded by PRIN 2022 PNRR, MSCA Actions

(7/7)

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

All details can be found in our paper.

📄 arXiv: arxiv.org/abs/2507.16015
🌐 Webpage: machinelearning.uniud.it/datasets/vista

The VISTA benchmark will be released soon. Stay tuned!

(6/7)

Is Tracking really more challenging in First Person Egocentric Vision?

Visual object tracking and segmentation are becoming fundamental tasks for understanding human activities in egocentric vision. Recent research has benchmarked state-of-the-art methods and concluded t...

arxiv.org

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

These FPV-specific challenges include:
- Frequent object disappearances
- Continuous camera motion altering object appearance
- Object distractors
- Wide field-of-view distortions near frame edges

(5/7)

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

- Trackers learn viewpoint biases and perform best on the viewpoint used during training.
- FPV tracking presents its specific challenges.

(4/7)

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

Key takeaways from our study:

- FPV is challenging for state-of-the-art generalistic trackers.
- Tracking objects in human-object interaction videos is difficult across both first- and third-person viewpoints.

(3/7)

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

We specifically examined whether these drops are due to FPV itself or to the complexity of human-object interaction scenarios.

To do this, we designed VISTA, a benchmark using synchronized first and third-person recordings of the same activities.

(2/7)

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Jul 23

Is Tracking really more challenging in First Person Egocentric Vision?

Our new #ICCV2025 paper follows up our IJCV 2023 study (bit.ly/4nVRJw9).

We further investigate the causes of performance drops in object and segmentation tracking under egocentric FPV settings.

🧵 (1/7)

1 1

Matteo Dunnhofer @mdunnhofer.bsky.social · Jun 20

In the past weeks we have been around NETI at @utaustin.bsky.social, @vssmtg.bsky.social 2025, and the CVR-CIAN conf. @yorkuniversity.bsky.social, to discuss early findings on modeling object motion in the macaque visual system by deep neural networks. Details to appear soon! @kohitij.bsky.social

2

Matteo Dunnhofer @mdunnhofer.bsky.social · May 29

Details: sciencedirect.com/special-issu...

Submissions open: June 1, 2025
Deadline: September 30, 2025

2/2

CVIU | Computer Vision and Image Understanding | ScienceDirect.com by Elsevier - Computer Vision and Image Understanding | ScienceDirect.com by ElsevierScienceDirect

Read the latest articles of Computer Vision and Image Understanding at ScienceDirect.com, Elsevier’s leading platform of peer-reviewed scholarly literature

sciencedirect.com

Matteo Dunnhofer @mdunnhofer.bsky.social · May 29

The teams behind the workshops on Computer Vision for Winter Sports at @wacvconference.bsky.social and on Computer Vision in Sports at @cvprconference.bsky.social have joined forces in the organisation of a special issue in CVIU on topics related to computer vision applications in sports.

1/2

1 1 1

Matteo Dunnhofer @mdunnhofer.bsky.social · May 10

Honored to be on the list this year!

#CVPR2026 @cvprconference.bsky.social · May 10

Behind every great conference is a team of dedicated reviewers. Congratulations to this year’s #CVPR2025 Outstanding Reviewers!

cvpr.thecvf.com/Conferences/...

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 4

We are now live with the 3d Workshop on Computer Vision for Winter Sports @wacvconference.bsky.social #WACV2025.

Make sure to attend if you are around!

Matteo Dunnhofer @mdunnhofer.bsky.social · Dec 1

The #SkiTB Visual Tracking Challenge at #WACV2025 is open for submissions!

The goal is to track a skier in a video capturing his/her full performance across multiple video cameras, and it is based on our recently released SkiTB dataset.

🎥 youtu.be/Aos5iKrYM5o

1/3

[WACV 2024] Tracking Skiers from the Top to the Bottom - qualitative examples

YouTube video by Matteo Dunnhofer

youtu.be

1 2

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 3

This paper contributes to our projects PRIN 2022 EXTRA EYE and Project PRIN 2022 PNRR TEAM funded by European Union-NextGenerationEU.

6/6

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 3

This work was led by Moritz Nottebaum (stop by his poster!) at the Machine Learning and Perception Lab of the University of Udine

5/6

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 3

LowFormer achieves significant speedups in image throughput and latency on various hardware platforms, while maintaining or surpassing the accuracy of current state-of-the-art models across image recognition, object detection, and semantic segmentation.

4/6

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 3

We used insights from such an analysis to enhance the hardware efficiency of backbones at the macro level, and introduced a slimmed-down version of multi-head self-attention to improve efficiency in the micro design.

3/6

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 3

We empirically found out that MACs alone do not accurately account for inference speed.

2/6

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Mar 3

Today at @wacvconference.bsky.social #WACV2025, we present LowFormer, a new family of convolutional-transformer architectures for computer vision that improve efficiency by optimizing running time on hardware rather than minimizing MAC operations.

arXiv: arxiv.org/abs/2409.03460

1/6

LowFormer: Hardware Efficient Design for Convolutional Transformer Backbones

Research in efficient vision backbones is evolving into models that are a mixture of convolutions and transformer blocks. A smart combination of both, architecture-wise and component-wise is mandatory...

arxiv.org

1 1 3

Matteo Dunnhofer @mdunnhofer.bsky.social · Dec 23

Is attendance open to YorkU researchers (e.g. postdocs)? Would like a lot to learn from your teaching style!

1 3

Matteo Dunnhofer @mdunnhofer.bsky.social · Dec 15

Did the same a few weeks ago in Toronto. I think this is the best pizza flavor you can get in Canada 😂

1 1

Matteo Dunnhofer @mdunnhofer.bsky.social · Dec 1

The top-performing teams will be invited to present their solution at the 3rd Workshop on Computer Vision for Winter Sports at #WACV2025!

📄 sites.google.com/unitn.it/cv4...

3/3

CV4WS@WACV2025

UPDATE (11/19/24): DEADLINE EXTENDED

sites.google.com

1

Matteo Dunnhofer @mdunnhofer.bsky.social · Dec 1

The challenge platform is hosted on CodaLab and you can find all the submission instructions there.

Deadline for submission is January 31st 2025.

🏆 codalab.lisn.upsaclay.fr/competitions...

2/3

CodaLab - Competition

codalab.lisn.upsaclay.fr

1 1

Matteo Dunnhofer @mdunnhofer.bsky.social · Dec 1

The #SkiTB Visual Tracking Challenge at #WACV2025 is open for submissions!

The goal is to track a skier in a video capturing his/her full performance across multiple video cameras, and it is based on our recently released SkiTB dataset.

🎥 youtu.be/Aos5iKrYM5o

1/3

[WACV 2024] Tracking Skiers from the Top to the Bottom - qualitative examples

YouTube video by Matteo Dunnhofer

youtu.be

1 1 5

Matteo Dunnhofer @mdunnhofer.bsky.social · Nov 24

Would be happy to be added as well :) I am working on visual tracking, currently at YorkU ;)