[FEATURE_REQUEST] Adding the 2020 SOTA paper for image retrieval (similarity scenario)
mhezarei opened this issue · comments
Description
The current SOTA paper for image retrieval was published in 2020. It's called "Combination of Multiple Global Descriptors for Image Retrieval," and you can find it here.
Here's what the results table would look like with the addition of this paper:
Paper | Year | Uses triplet learning | Recall@1 CARS196 | Recall@1 CUB200-2011 | Recall@1 SOP |
---|---|---|---|---|---|
Deep Metric Learning via Lifted Structured Feature Embedding | CVPR 2016 | 49% | 47% | 62% | |
Deep Metric learning with angular loss | ICCV 2017 | Yes | 71% | 55% | 71% |
Sampling Matters in Deep Embedding Learning | ICCV 2017 | Yes | 80% | 64% | 73% |
No Fuss Distance Metric Learning using Proxies | ICCV 2017 | Yes | 73% | 49% | 74% |
Deep metric learning with hierarchical triplet loss | ECCV 2018 | Yes | 81% | 57% | 75% |
Classification is a Strong Baseline for DeepMetric Learning (Implemented in this repository) |
BMVC 2019 | No | 84% (512-dim) 89% (2048-dim) |
61% (512-dim) 65% (2048-dim) |
78% (512-dim) 80% (2048-dim) |
Combination of Multiple Global Descriptors for Image Retrieval | 2020 | Yes | 94.8% (1536-dim) | 79.2% (1536-dim) | 84.2% (1536-dim) |
Although the paper uses the triplet loss (which is somewhat against the idea of this section), I believe the improvements are so considerable that it is worth ignoring that fact.
Also, the implementation code is available here, but it needs a few adjustments to be suitable for this repo. I was wondering if I could be helpful by doing so.
Expected behavior with the suggested feature
Other Comments
Thanks for the suggestion. Next time we update the code/table we will also take this reference into account.