Task Indicating Transformer for Task-conditional Dense Predictions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lu, Yuxiang, Sirejiding, Shalayiding, Bayramli, Bayram, Huang, Suizhi, Ding, Yue, Lu, Hongtao
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929261132120064
author Lu, Yuxiang
Sirejiding, Shalayiding
Bayramli, Bayram
Huang, Suizhi
Ding, Yue
Lu, Hongtao
author_facet Lu, Yuxiang
Sirejiding, Shalayiding
Bayramli, Bayram
Huang, Suizhi
Ding, Yue
Lu, Hongtao
contents The task-conditional model is a distinctive stream for efficient multi-task learning. Existing works encounter a critical limitation in learning task-agnostic and task-specific representations, primarily due to shortcomings in global context modeling arising from CNN-based architectures, as well as a deficiency in multi-scale feature interaction within the decoder. In this paper, we introduce a novel task-conditional framework called Task Indicating Transformer (TIT) to tackle this challenge. Our approach designs a Mix Task Adapter module within the transformer block, which incorporates a Task Indicating Matrix through matrix decomposition, thereby enhancing long-range dependency modeling and parameter-efficient feature adaptation by capturing intra- and inter-task features. Moreover, we propose a Task Gate Decoder module that harnesses a Task Indicating Vector and gating mechanism to facilitate adaptive multi-scale feature refinement guided by task embeddings. Experiments on two public multi-task dense prediction benchmarks, NYUD-v2 and PASCAL-Context, demonstrate that our approach surpasses state-of-the-art task-conditional methods.
format Preprint
id arxiv_https___arxiv_org_abs_2403_00327
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Task Indicating Transformer for Task-conditional Dense Predictions
Lu, Yuxiang
Sirejiding, Shalayiding
Bayramli, Bayram
Huang, Suizhi
Ding, Yue
Lu, Hongtao
Computer Vision and Pattern Recognition
The task-conditional model is a distinctive stream for efficient multi-task learning. Existing works encounter a critical limitation in learning task-agnostic and task-specific representations, primarily due to shortcomings in global context modeling arising from CNN-based architectures, as well as a deficiency in multi-scale feature interaction within the decoder. In this paper, we introduce a novel task-conditional framework called Task Indicating Transformer (TIT) to tackle this challenge. Our approach designs a Mix Task Adapter module within the transformer block, which incorporates a Task Indicating Matrix through matrix decomposition, thereby enhancing long-range dependency modeling and parameter-efficient feature adaptation by capturing intra- and inter-task features. Moreover, we propose a Task Gate Decoder module that harnesses a Task Indicating Vector and gating mechanism to facilitate adaptive multi-scale feature refinement guided by task embeddings. Experiments on two public multi-task dense prediction benchmarks, NYUD-v2 and PASCAL-Context, demonstrate that our approach surpasses state-of-the-art task-conditional methods.
title Task Indicating Transformer for Task-conditional Dense Predictions
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.00327