Name
Abstract | ||
|---|---|---|
This paper presents a method to find a photo set of a given event by avoiding misrecognitions from neighbor events. In contrast with a photo summarization problem that seeks photos with high coverage for a photo collection of an event, our goal is to find photos with high perceptual quality, which is defined as the accuracy and quickness to recognize an certain event. We take an approach to proactively avoid misrecognitions of a photo set to improve its perceptual quality, as a photo set that cannot be misrecognized as neighbor events is expected to be one with high perceptual quality. We discuss the types of misrecognitions, namely, sub-event, super-event and sibling-event misrecognition, and then propose three criteria corresponding to each of them. By maximizing combination of three criteria, we generate a photo set that minimizes misrecognition against neighbor events. We empirically demonstrated that our proposed approach can generate photo sets with high perceptual quality in comparison with a baseline method. |
Year | Venue | Keywords |
|---|---|---|
2014 | WEB-AGE INFORMATION MANAGEMENT, WAIM 2014 | Photo Set, Event, Misrecognition |
Field | DocType | Volume |
Automatic summarization,Computer science,Artificial intelligence,Perception,Machine learning | Conference | 8485 |
ISSN | Citations | PageRank |
0302-9743 | 0 | 0.34 |
References | Authors | |
0 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Bei Liu | 1 | 26 | 12.94 |
| Makoto P. Kato | 2 | 138 | 19.24 |
| Katsumi Tanaka | 3 | 0 | 0.34 |