Using only the training data for the new classes, we pro- pose an incremental instance segmentation method, which utilizes multi-teacher networks to supervise ...

May 18, 2021 · We propose an incremental instance segmentation method that consists of three networks: Former Teacher Network (FTN), Current Student Network (CSN) and Current ...

Code for AAAI2021 paper "Class-Incremental Instance Segmentation via Multi-Teacher Networks". 8 stars 2 forks Branches Tags Activity.

Abstract: Although deep neural networks have achieved amazing results on instance segmentation, they are still ill-equipped when they are required to learn ...

To address these problems, we propose an incremental instance segmentation method that consists of three networks: Former Teacher Network (FTN), Current Student ...

Class-incremental instance segmentation via multi-teacher networks. Y Gu, C Deng, K Wei. Proceedings of the AAAI Conference on Artificial Intelligence 35 (2) ...

Code for AAAI2021 paper "Class-Incremental Instance Segmentation via Multi-Teacher Networks". Python 8 2 · DBPN DBPN Public. Deep Back-Projection Networks For ...

This paper aims to attain CL in a more effort-saving sce- nario for IS, where our model learns to predict instance- level segmentation using weakly supervised ...

Class-Incremental Instance Segmentation via Multi-Teacher Networks [PDF][Code] ... Multi-Class Support Vector Machine via Maximizing Multi-Class Margins [PDF].

Class-incremental instance segmentation via multi-teacher networks. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 1478–1486, 2021 ...