UTAR students clinch first place at ICIP 2025 global AI challenge

UTAR students Wong Yi Jie (team leader) and Tan Jing Jie, both PhD candidates in Engineering at the Lee Kong Chian Faculty of Engineering and Science (LKC FES), have secured first place at the Cityscape Aerial Image Dataset for Object Detection Challenge (CADOT 2025), held on 17 September 2025 in Anchorage, Alaska, United States. The competition was part of the IEEE International Conference on Image Processing (ICIP 2025), which took place from 14 to 18 September 2025 at the same venue. ICIP is the world's largest and most comprehensive technical conference focused on image and video processing and computer vision. 

Organised by ICIP in collaboration with Université Sorbonne Paris Nord (USPN), Laboratoire de Traitement et Transport de l’Information (L2TI) and COSE, a French high-tech SME, the CADOT Challenge brought together global researchers to tackle the complexities of object detection in dense urban environments using high-resolution aerial imagery. Participants were tasked with designing learning-based models capable of interpreting intricate cityscapes and supporting AI-driven solutions for urban-scale applications. 

Under the guidance of their supervisors, Assoc Prof Ir Ts Dr Tham Mau Luen, Assoc Prof Dr Kwan Ban Hoe and Assoc Prof Ir Ts Dr Hum Yan Chai, the UTAR duo presented their work titled Tri-Axial Scaling in Aerial Object Detection: Model Size, Dataset Size and Quality, and Test-Time Inference in the CADOT Challenge. Drawing inspiration from scaling laws and test-time adaptation techniques from large language models, their framework effectively balanced model scaling, dataset quality and test-time inference, ultimately setting a new benchmark in aerial object detection research. 

Although Wong and Tan did not travel to Alaska for the conference, their research stood out amongst international competitors, earning them the championship title and a cash prize of €1,000 alongside certificates of recognition. 

Reflecting on the victory, Wong shared, “We joined the competition to challenge ourselves in a global academic setting and to test our skills in AI beyond the scope of our research. Winning this championship is deeply meaningful as it validates our contributions and motivates us to continue pushing the boundaries of our work.” 

Tan echoed this sentiment, adding, “This is my first global academic championship, and I am truly honoured. It is an important milestone in my journey as a researcher, inspiring me to make a broader contribution to the field.”