MST: Adaptive Multi-Scale Tokens Guided Interactive Segmentation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xu, Long, Li, Shanghong, Chen, Yongquan, Luo, Jun, Lai, Shiwu
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866929232299425792
author Xu, Long
Li, Shanghong
Chen, Yongquan
Luo, Jun
Lai, Shiwu
author_facet Xu, Long
Li, Shanghong
Chen, Yongquan
Luo, Jun
Lai, Shiwu
contents Interactive segmentation has gained significant attention for its application in human-computer interaction and data annotation. To address the target scale variation issue in interactive segmentation, a novel multi-scale token adaptation algorithm is proposed. By performing top-k operations across multi-scale tokens, the computational complexity is greatly simplified while ensuring performance. To enhance the robustness of multi-scale token selection, we also propose a token learning algorithm based on contrastive loss. This algorithm can effectively improve the performance of multi-scale token adaptation. Extensive benchmarking shows that the algorithm achieves state-of-the-art (SOTA) performance, compared to current methods. An interactive demo and all reproducible codes will be released at https://github.com/hahamyt/mst.
format Preprint
id arxiv_https___arxiv_org_abs_2401_04403
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MST: Adaptive Multi-Scale Tokens Guided Interactive Segmentation
Xu, Long
Li, Shanghong
Chen, Yongquan
Luo, Jun
Lai, Shiwu
Computer Vision and Pattern Recognition
Interactive segmentation has gained significant attention for its application in human-computer interaction and data annotation. To address the target scale variation issue in interactive segmentation, a novel multi-scale token adaptation algorithm is proposed. By performing top-k operations across multi-scale tokens, the computational complexity is greatly simplified while ensuring performance. To enhance the robustness of multi-scale token selection, we also propose a token learning algorithm based on contrastive loss. This algorithm can effectively improve the performance of multi-scale token adaptation. Extensive benchmarking shows that the algorithm achieves state-of-the-art (SOTA) performance, compared to current methods. An interactive demo and all reproducible codes will be released at https://github.com/hahamyt/mst.
title MST: Adaptive Multi-Scale Tokens Guided Interactive Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2401.04403