Task-Guided Multi-Annotation Triplet Learning for Remote Sensing Representations
For remote sensing applications requiring multi-task learning with multiple annotations, this method improves shared representation quality without manual weight tuning.
The paper proposes a task-guided multi-annotation triplet loss that uses mutual-information criteria to select informative triplets across tasks, eliminating the need for static weight tuning. Experiments on an aerial wildlife dataset show improved classification and regression performance compared to prior triplet loss methods.
Prior multi-task triplet loss methods relied on static weights to balance supervision between various types of annotation. However, static weighting requires tuning and does not account for how tasks interact when shaping a shared representation. To address this, the proposed task-guided multi-annotation triplet loss removes this dependency by selecting triplets through a mutual-information criteria that identifies triplets most informative across tasks. This strategy modifies which samples influence the representation rather than adjusting loss magnitudes. Experiments on an aerial wildlife dataset compare the proposed task-guided selection against several triplet loss setups for shaping a representation in an effective multi-task manner. The results show improved classification and regression performance and demonstrate that task-aware triplet selection produces a more effective shared representation for downstream tasks.