RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations

TL;DR

RELOCATE is a training-free, region-based method for visual query localization in long videos, leveraging pretrained models and enhancements to outperform prior methods significantly.

Contribution

It introduces a simple, training-free baseline using region-based representations and novel refinements for effective visual query localization in videos.

Findings

Outperforms prior methods by 49% in average precision

Effective in handling small objects and cluttered scenes

Establishes a new baseline on Ego4D dataset

Abstract

We present RELOCATE, a simple training-free baseline designed to perform the challenging task of visual query localization in long videos. To eliminate the need for task-specific training and efficiently handle long videos, RELOCATE leverages a region-based representation derived from pretrained vision models. At a high level, it follows the classic object localization approach: (1) identify all objects in each video frame, (2) compare the objects with the given query and select the most similar ones, and (3) perform bidirectional tracking to get a spatio-temporal response. However, we propose some key enhancements to handle small objects, cluttered scenes, partial visibility, and varying appearances. Notably, we refine the selected objects for accurate localization and generate additional visual queries to capture visual variations. We evaluate RELOCATE on the challenging Ego4D Visual Query 2D Localization dataset, establishing a new baseline that outperforms prior task-specific methods by 49% (relative improvement) in spatio-temporal average precision.