Dear MARVEL community,

Welcome to the 7th edition of our MARVEL newsletter! As we approach the summer break, we are excited to share with you the culmination of a period of intense work and significant progress in our project. This edition is packed with the results and outcomes we've achieved recently, and it offers a glimpse of what's to come in future updates.

Since our last newsletter, we've been busy participating in a series of events. We've made our mark at WACV 2023, HiPEAC2023, Vision Zero for the Balkans Road Safety Conference 2023, and Data Week 2023. These platforms have allowed us to showcase our progress and explore potential collaborations.

Our blog section is richer than ever, featuring posts that delve into various aspects of our project. From the advantages of the MARVEL framework as perceived by end-users in urban security, to the growing community of TinyML, the power of edge computing, modeling emotions, and the fascinating link between MEMS microphones and the digital world, there's a wealth of information waiting for you.

We're also excited to share our latest YouTube video, offering a deeper understanding of the MARVEL project. It's a must-watch for anyone interested in our work.

The "Deep Learning for Autonomous Systems and Smart Cities" summer school event was organized by Aarhus University and ran from May 23 to May 26, 2023. This event was a unique intersection of two H2020 Research and Innovation projects, MARVEL and OpenDR, which jointly organized the summer school. The focus of the summer school was on the application of machine learning, specifically deep learning, in autonomous systems and smart cities.

Visit our zenodo community and get access to a great number of scientific papers, presentations, datasets, and deliverable reports. The outcomes of our latest research include:

• Audio-Visual Dataset and Method for Anomaly Detection in Traffic Videos
• Learning distinct features helps, provably,
• Crowd Counting on Heavily Compressed Images with Curriculum Pre-Training,
• Efficient High-Resolution Deep Learning: A Survey,
• Accurate Gigapixel Crowd Counting by Iterative Zooming and Refinement
  
• PromptMix: Text-to-image diffusion models enhance the performance of lightweight networks.

We're excited to share with you a special treat! Back in April, our dedicated technical team at MARVEL conducted a live demonstration showcasing the capabilities of our innovative framework. We've captured the magic of that event and it's now available for you to experience on our YouTube channel. Watch the video on our YouTube channel by clicking the image below: 

Your thoughts, feedback, and ideas are invaluable to us. After watching the demonstration, we invite you to share your impressions and thoughts. Did you find the demonstration insightful? Do you have questions about what you saw? Or perhaps you have suggestions on how we can improve?

Reach out to us through our contact page: Contact Us

Your voice matters to us. By sharing your thoughts, you're helping us shape the future of MARVEL. We look forward to hearing from you!