Learning Regional Attention Over Multi-Resolution Deep Convolutional Features For Trademark Retrieval

Abstract

Large-scale trademark retrieval is an important content-based image retrieval task. A recent study shows that off-theshelf deep features aggregated with Regional-Maximum Activation of Convolutions (R-MAC) achieve state-of-theart results. However, R-MAC suffers in the presence of background clutter/trivial regions and scale variance, and discards important spatial information. We introduce three simple but effective modifications to R-MAC to overcome these drawbacks. First, we propose the use of both sum and max pooling to minimise the loss of spatial information. We also employ domain-specific unsupervised soft-attention to eliminate background clutter and unimportant regions. Finally, we add multi-resolution inputs to enhance the scaleinvariance of R-MAC. We evaluate these three modifications on the million-scale METU dataset. Our results show that all modifications bring non-trivial improvements, and surpass previous state-of-the-art results.

Trademark Similarity Evaluation Using a Combination of ViT and Local Features

“…In [15], improvements of R-MAC such as MR (Multi-Resolution), SMAC (sum and max activation of convolution), and URA (unsupervised regional attention) were considered. This work shows state-of-the-art results with a mAP of 31.0% and NAR of 0.028.…”