Beyond Resolution: Object-to-Image Ratio Mismatch in Instance Retrieval
What it is
This paper identifies object-to-image (O2I) ratio mismatch, rather than resolution loss, as the primary cause of failure in visual instance retrieval when object sizes differ between query and gallery. It proposes query-side scale augmentation and a crop reranker to mitigate this issue, achieving state-of-the-art results on the ILIAS 100M benchmark without modifying the precomputed gallery index.
The problem it solves
Visual instance retrieval fails when the same object appears at different apparent sizes (different fractions of the image area) in the query versus the gallery, a problem previously misattributed to resolution loss.
What you could use it for
- Improving accuracy in large-scale visual instance retrieval systems
- Enhancing robustness of object detection models to varying object scales
- Optimizing search performance in image databases with inconsistent object framing
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.