Research Article
PTF-SimCM: A Simple Contrastive Model with Polysemous Text Fusion for Visual Similarity Metric
Table 5
Accuracy on Pascal Sentence.
and
denote the number of embeddings and dimensions of metric embedding, respectively.
| Methods | Recall@k | R-Precision | MAP@R | k = 1 | k = 2 | k = 4 | k = 8 |
| Siamese | 0.581 | 0.709 | 0.784 | 0.860 | 0.326 | 0.218 | Triplet | 0.560 | 0.671 | 0.773 | 0.842 | 0.318 | 0.202 | SoftTriple | 0.598 | 0.712 | 0.802 | 0.864 | 0.343 | 0.228 | Label Relaxation | 0.579 | 0.687 | 0.785 | 0.849 | 0.308 | 0.198 | MemVir | 0.593 | 0.709 | 0.796 | 0.876 | 0.337 | 0.232 | SimSiam | 0.443 | 0.585 | 0.716 | 0.821 | 0.259 | 0.147 | SimSiam + proj | 0.491 | 0.606 | 0.702 | 0.789 | 0.217 | 0.138 | Ours () | 0.592 | 0.714 | 0.806 | 0.864 | 0.336 | 0.240 | Ours () | 0.616 | 0.719 | 0.801 | 0.870 | 0.299 | 0.206 |
|
|