I arrived in Dalian in the first week of June to begin a three-month academic placement at the Cyber-Physical Systems Centre, Dalian University of Technology (DUT).
In the first days, I met Professor Liming Chen and his team, joined their activities, and quickly adapted to the new city and the research environment. Their warm welcome made it easy to start focusing on my project right away.
DUT provided excellent working conditions, with a dedicated office, access to powerful GPU servers, and continuous support from the research team. They assisted me in setting up experiments, solving technical issues, and improving efficiency on large-scale servers. Through regular discussions, I gained fresh perspectives on multimodal AI, and our joint problem-solving sessions strengthened collaboration within the group. Working alongside such a supportive and talented team was one of the most rewarding aspects of my placement.
With colleagues at DUT
With guidance from Professor Chen and my UCD supervisor, Professor Tahar Kechadi, I followed a clear research process. I first developed a model that could detect objects in complex scenes. I then expanded the system to also track objects over time. Finally, I added short-term trajectory prediction, enabling the model to forecast how objects would move in dynamic environments.
One of the main challenges I faced was scale variation, as objects moved closer or further from the camera. With advice from Professor Chen, support from Professor Tahar, and input from the DUT team, I explored multi-scale feature representations to make the tracker more robust. Through this iterative process, I refined the framework into a unified, lightweight transformer-based system that is privacy-preserving, GPU-efficient, and deployment-ready—capable of handling occlusions, rapid motion, and varied environments.
On 23rd August, I assisted at the HMCC & HHME Conference 2025 in Dalian—one of the largest and most valuable conferences in computer science—chaired by Professor Chen and hosted by DUT. At his invitation, I joined the organizing team as a volunteer, working closely with colleagues to support sessions and coordinate with international researchers. Collaborating with the conference team was a valuable learning experience that allowed me to contribute directly to the success of an international event, while also gaining exposure to the global AI community and seeing DUT’s leading role in advancing international research collaboration.
With the organizing team at HMCC & HHME 2025 in Dalian
Life in Dalian was a truly inspiring experience. I loved the city’s warm atmosphere, the chance to visit beautiful places such as Xinghai Square and Tiger Beach, and the opportunity to enjoy its famous seafood. Most of all, I was touched by the kindness and hospitality of the people, which made me feel welcome and at home throughout my stay.
This placement strengthened my research, enriched my cultural experience, and opened new opportunities for collaboration between ML-Labs and DUT. It has shaped my current work and will serve as a foundation for future partnerships in multimodal AI and autonomous systems.
I am deeply grateful to ML-Labs for supporting this opportunity and to DUT for providing such a productive and inspiring environment.
— Zhongping Dong
ML-Labs Student Cohort 5