Bootstrapping Video Semantic Segmentation Model via Distillation-assisted Test-Time Adaptation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kim, Jihun, Kwon, Hoyong, Kweon, Hyeokjun, Yoon, Kuk-Jin
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914470622658560
author Kim, Jihun
Kwon, Hoyong
Kweon, Hyeokjun
Yoon, Kuk-Jin
author_facet Kim, Jihun
Kwon, Hoyong
Kweon, Hyeokjun
Yoon, Kuk-Jin
contents Fully supervised Video Semantic Segmentation (VSS) relies heavily on densely annotated video data, limiting practical applicability. Alternatively, applying pre-trained Image Semantic Segmentation (ISS) models frame-by-frame avoids annotation costs but ignores crucial temporal coherence. Recent foundation models such as SAM2 enable high-quality mask propagation yet remain impractical for direct VSS due to limited semantic understanding and computational overhead. In this paper, we propose DiTTA (Distillation-assisted Test-Time Adaptation), a novel framework that converts an ISS model into a temporally-aware VSS model through efficient test-time adaptation (TTA), without annotated videos. DiTTA distills SAM2's temporal segmentation knowledge into the ISS model during a brief, single-pass initialization phase, complemented by a lightweight temporal fusion module to aggregate cross-frame context. Crucially, DiTTA achieves robust generalization even when adapting with highly limited partial video snippets (e.g., initial 10%), significantly outperforming zero-shot refinement approaches that repeatedly invoke SAM2 during inference. Extensive experiments on VSPW and Cityscapes demonstrate DiTTA's effectiveness, achieving competitive or superior performance relative to fully-supervised VSS methods, thus providing a practical and annotation-free solution for real-world VSS tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2604_10950
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Bootstrapping Video Semantic Segmentation Model via Distillation-assisted Test-Time Adaptation
Kim, Jihun
Kwon, Hoyong
Kweon, Hyeokjun
Yoon, Kuk-Jin
Computer Vision and Pattern Recognition
Fully supervised Video Semantic Segmentation (VSS) relies heavily on densely annotated video data, limiting practical applicability. Alternatively, applying pre-trained Image Semantic Segmentation (ISS) models frame-by-frame avoids annotation costs but ignores crucial temporal coherence. Recent foundation models such as SAM2 enable high-quality mask propagation yet remain impractical for direct VSS due to limited semantic understanding and computational overhead. In this paper, we propose DiTTA (Distillation-assisted Test-Time Adaptation), a novel framework that converts an ISS model into a temporally-aware VSS model through efficient test-time adaptation (TTA), without annotated videos. DiTTA distills SAM2's temporal segmentation knowledge into the ISS model during a brief, single-pass initialization phase, complemented by a lightweight temporal fusion module to aggregate cross-frame context. Crucially, DiTTA achieves robust generalization even when adapting with highly limited partial video snippets (e.g., initial 10%), significantly outperforming zero-shot refinement approaches that repeatedly invoke SAM2 during inference. Extensive experiments on VSPW and Cityscapes demonstrate DiTTA's effectiveness, achieving competitive or superior performance relative to fully-supervised VSS methods, thus providing a practical and annotation-free solution for real-world VSS tasks.
title Bootstrapping Video Semantic Segmentation Model via Distillation-assisted Test-Time Adaptation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.10950