GPR1200: A Benchmark for General-Purpose Content-Based Image Retrieval
This addresses the problem for researchers and practitioners needing to benchmark CBIR models across diverse domains, though it is incremental as it builds on existing retrieval training methods.
The authors tackled the lack of a benchmark for evaluating general-purpose content-based image retrieval (CBIR) models by introducing GPR1200, a manually curated dataset with a broad range of image categories, and showed that large-scale pretraining significantly improves retrieval performance, with experiments on fine-tuning to further enhance these properties.
Even though it has extensively been shown that retrieval specific training of deep neural networks is beneficial for nearest neighbor image search quality, most of these models are trained and tested in the domain of landmarks images. However, some applications use images from various other domains and therefore need a network with good generalization properties - a general-purpose CBIR model. To the best of our knowledge, no testing protocol has so far been introduced to benchmark models with respect to general image retrieval quality. After analyzing popular image retrieval test sets we decided to manually curate GPR1200, an easy to use and accessible but challenging benchmark dataset with a broad range of image categories. This benchmark is subsequently used to evaluate various pretrained models of different architectures on their generalization qualities. We show that large-scale pretraining significantly improves retrieval performance and present experiments on how to further increase these properties by appropriate fine-tuning. With these promising results, we hope to increase interest in the research topic of general-purpose CBIR.