Multi-Task Learning with Multi-Annotation Triplet Loss for Improved Object Detection
This work addresses a specific limitation in multi-task learning for object detection, offering an incremental improvement for researchers and practitioners in computer vision.
The paper tackled the problem of triplet loss not using all available annotations in multi-task learning by introducing a Multi-Annotation Triplet Loss (MATL) framework that incorporates additional annotations like bounding box information, resulting in improved performance in classification and localization on an aerial wildlife imagery dataset.
Triplet loss traditionally relies only on class labels and does not use all available information in multi-task scenarios where multiple types of annotations are available. This paper introduces a Multi-Annotation Triplet Loss (MATL) framework that extends triplet loss by incorporating additional annotations, such as bounding box information, alongside class labels in the loss formulation. By using these complementary annotations, MATL improves multi-task learning for tasks requiring both classification and localization. Experiments on an aerial wildlife imagery dataset demonstrate that MATL outperforms conventional triplet loss in both classification and localization. These findings highlight the benefit of using all available annotations for triplet loss in multi-task learning frameworks.