🤖 AI Summary
To address the challenge of rapid memory retrieval from large-scale, time-dense, and semantically sparse lifelog data, this paper proposes an interactive retrieval framework integrating time-aware querying with composable concept filtering. Methodologically, it introduces a lightweight timeline browsing interface, enhanced daily summary generation, and multi-granularity semantic filters—dynamically combinable across attributes, activities, and objects—alongside an optimized temporal index enabling millisecond-level response times. The key contributions are: (1) explicit modeling of temporal constraints as retrieval priors, and (2) daily-summary-driven joint alignment of concepts and timestamps. Evaluated on the Lifelog Search Challenge 2021 benchmark, the system achieves top-3 average accuracy and the highest interaction efficiency within strict time limits, demonstrating both practical utility and technical competitiveness in real-world lifelog retrieval scenarios.
📝 Abstract
Since its first iteration in 2018, the Lifelog Search Challenge (LSC) continues to rise in popularity as an interactive lifelog data retrieval competition, co-located at the ACM International Conference on Multimedia Retrieval (ICMR). The goal of this annual live event is to search a large corpus of lifelogging data for specifically announced memories using a purposefully developed tool within a limited amount of time. As long-standing participants, we present our improved lifeXplore - a retrieval system combining chronologic day summary browsing with interactive combinable concept filtering. Compared to previous versions, the tool is improved by incorporating temporal queries, advanced day summary features as well as usability improvements.