This document summarizes an approach for diversifying image retrieval results using adaptive clustering. It explores using various visual and textual features with clustering algorithms like k-means and affinity propagation. Evaluation on a benchmark task shows the best approach achieves a precision of 0.8250, recall of 0.6634, and F1-score of 0.7186. While fixed settings underperform, the adaptive approach achieves a precision of 0.6500, recall of 0.4398, and F1-score of 0.5123 on visual features alone. Lessons learned are that different queries favor different features, approaches need flexibility, and the evaluation could be made more objective.
MediaEval 2016: An adaptive clustering approach for the diversification of image retrieval results
1. AN ADAPTIVE CLUSTERING APPROACH FOR THE
DIVERSIFICATION OF IMAGE RETRIEVAL RESULTS
MAIA ZAHARIEVA
VIENNA UNIVERSITY OF TECHNOLOGY & UNIVERSITY OF VIENNA, AUSTRIA
MAIA.ZAHARIEVA@TUWIEN.AC.AT
October 20-21, Hilversum, Netherlands
RETRIEVING DIVERSE SOCIAL IMAGES TASK
13. MEDIAEVAL BENCHMARK 2016: RETRIEVING DIVERSE SOCIAL IMAGES TASK
LESSONS LEARNED
▸ Different queries favour different feature representations
▸ Strong requirements for higher generalization applicability
and flexibility of approaches for image search diversification
▸ Multiple solutions can be considered being correct
▸ How to make the evaluation more objective?