Make One-Shot Video Object Segmentation Efficient Again

TL;DR

This paper introduces e-OSVOS, an efficient one-shot video object segmentation method that decouples detection and segmentation, uses meta-learned initialization and learning rates, and applies online adaptation to improve speed and accuracy.

Contribution

It proposes a novel, efficient VOS approach that optimizes test-time performance through meta-learning and online adaptation, reducing runtime while maintaining state-of-the-art accuracy.

Findings

Achieves state-of-the-art results on DAVIS and YouTube-VOS datasets.

Significantly reduces test runtime compared to previous methods.

Maintains high segmentation accuracy with online model adaptation.

Abstract

Video object segmentation (VOS) describes the task of segmenting a set of objects in each frame of a video. In the semi-supervised setting, the first mask of each object is provided at test time. Following the one-shot principle, fine-tuning VOS methods train a segmentation model separately on each given object mask. However, recently the VOS community has deemed such a test time optimization and its impact on the test runtime as unfeasible. To mitigate the inefficiencies of previous fine-tuning approaches, we present efficient One-Shot Video Object Segmentation (e-OSVOS). In contrast to most VOS approaches, e-OSVOS decouples the object detection task and predicts only local segmentation masks by applying a modified version of Mask R-CNN. The one-shot test runtime and performance are optimized without a laborious and handcrafted hyperparameter search. To this end, we meta learn the model initialization and learning rates for the test time optimization. To achieve optimal learning behavior, we predict individual learning rates at a neuron level. Furthermore, we apply an online adaptation to address the common performance degradation throughout a sequence by continuously fine-tuning the model on previous mask predictions supported by a frame-to-frame bounding box propagation. e-OSVOS provides state-of-the-art results on DAVIS 2016, DAVIS 2017, and YouTube-VOS for one-shot fine-tuning methods while reducing the test runtime substantially. Code is available at https://github.com/dvl-tum/e-osvos.