Combine knowledge of many vision transformers into a smaller multi-talented model
Combine knowledge of many vision transformers into a smaller multi-talented model
Knowledge Amalgamation for Object Detection with Transformers
arXiv paper abstract https://arxiv.org/abs/2203.03187v1
arXiv PDF paper https://arxiv.org/pdf/2203.03187v1.pdf
Knowledge amalgamation (KA) … aiming to transfer knowledge from several well-trained teachers to a multi-talented and compact student.
… In this work, … explore a more effective KA scheme for transformer-based object detection models.
… propose to dissolve the KA into two aspects: sequence-level amalgamation (SA) and task-level amalgamation (TA).
… a hint is generated within the sequence-level amalgamation by concatenating teacher sequences instead of redundantly aggregating them to a fixed-size one as previous KA works.
… the student learns heterogeneous detection tasks through soft targets with efficiency in the task-level amalgamation.
… experiments on PASCAL VOC and COCO … the sequence-level amalgamation significantly boosts the performance of students …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
