Introduction to the Fifth Annual Lifelog Search Challenge, LSC'22

Abstract

For the fifth time since 2018, the Lifelog Search Challenge (LSC) facilitated a benchmarking exercise to compare interactive search systems designed for multimodal lifelogs. LSC'22 attracted nine participating research groups who developed interactive lifelog retrieval systems enabling fast and effective access to lifelogs. The systems competed in front of a hybrid audience at the LSC workshop at ACM ICMR'22. This paper presents an introduction to the LSC workshop, the new (larger) dataset used in the competition, and introduces the participating lifelog search systems.

MEMORIA: A Memory Enhancement and MOment RetrIeval Application for LSC 2022

“…The First Workshop on Lifelogging Tools and Applications (LTA) in 2016 [10], aimed to discuss approaches to lifelog data capture, analytics and applications and to identify opportunities and challenges for researchers in this new and challenging area. Ever since, workshops and tasks have been pursued to advance research onto some of its key challenges: The Second Workshop on Lifelogging Tools and Applications (LTA 2017) [9], NTCIR Lifelog Tasks [13,14], Image-CLEFlifelog [3][4][5]28], and the Lifelogging Search Challenges (LSC) [15][16][17]. All these events contributed to the development of efficient interactive lifelog retrieval systems.…”

“…In this paper, we present the first prototype of MEMORIA (Memory Enhancement and MOment RetrIeval Application) for the participation at LSC challenge [17], with the aim of its evaluation and provide hints to improve our system in the future. Our system is divided into four main blocks, where users can upload their personal lifelogs and retrieve useful information using an interface based on keywords search and time periods.…”

Flexible Interactive Retrieval SysTem 3.0 for Visual Lifelog Exploration at LSC 2022

“…A system that can handle these new forms of data, which is more costly computational and storage-wise, can undoubtedly provide great value. While the LSC'22 dataset [8] is not a video dataset on its own, it is similar to one, in terms of size and temporal meaning. It is greater than the previous edition (roughly 725,000 images compared to 183,299 [7]) and poses a significant challenge to system developers [8].…”

“…While the LSC'22 dataset [8] is not a video dataset on its own, it is similar to one, in terms of size and temporal meaning. It is greater than the previous edition (roughly 725,000 images compared to 183,299 [7]) and poses a significant challenge to system developers [8].…”

Memento 2.0: An Improved Lifelog Search Engine for LSC'22

“…The Lifelog Search Challenge (LSC) [12] has attracted significant participation over the past 4 years since its inception in 2018. The LSC uses a multimodal dataset comprising egocentric images captured using a wearable camera apart from data coming through wearable sensors such as location, biometrics, sleep unlike other publicly available egocentric datasets which are generally unimodal and are mostly domain specific like EPIC-Kitchen [5] and EGO-CH [20].…”