We’re excited to share that our lab member Meilun Zhou has been awarded a travel grant from IEEE GRSS to attend IGARSS 2025 in Brisbane, Australia this August! Meilun’s paper was accepted for presentation at the conference, and the IEEE Geoscience and Remote Sensing Society recognized both the impact of the work and the importance of supporting early-career researchers through this competitive grant.

His paper proposes a Multi-Annotation Triplet Loss (MATL) framework that enhances traditional triplet loss by incorporating additional annotations—like bounding boxes—alongside class labels. By leveraging these complementary annotations, MATL improves both classification and localization performance in multi-task learning. Experiments on aerial wildlife imagery show that MATL outperforms standard triplet loss, demonstrating the value of using all available annotations.

For more information on the paper, click the following link: Multi-Task Learning with Multi-Annotation Triplet Loss for Improved Object Detection

Catch his presentation at IGARSS 2025 and learn how this approach is pushing the boundaries of object detection!

Congratulations, Meilun!