Research map / Image generation / Editing and style transfer
Unpaired Image Translation via Vector Symbolic Architectures
Justin Theiss, Jay Leverett, Daeil Kim, Aayush Prakash
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add VSA-based cyclic loss (ablation) | VSAIT · GTA to Cityscapes | mIoU: 10.21 -> 30.89 | Improved |
| learn the hypervector mapping with F instead of random mapping (ablation) | VSAIT · GTA to Cityscapes | mIoU: 17.53 -> 30.89 | Improved |
| use hypervector dimension 4096 instead of 128 (ablation) | VSAIT · GTA to Cityscapes | mIoU: 25.04 -> 30.89 | Improved |
| use hypervector dimension 2048 instead of 512 (ablation) | VSAIT · GTA to Cityscapes | mIoU: 29.35 -> 30.87 | Improved |
| increase VSA loss weight lambda from 1 to 10 (ablation) | VSAIT · GTA to Cityscapes | mIoU: 21.55 -> 30.89 | Improved |
| increase VSA loss weight lambda from 2 to 10 (ablation) | VSAIT · GTA to Cityscapes | mIoU: 26.49 -> 30.89 | Improved |
| increase lambda from 10 to 20 (ablation) | VSAIT · GTA to Cityscapes | mIoU: 30.89 -> 29.78 | Hurt |
| patch size 32 instead of 16 for feature extraction (ablation) | VSAIT · GTA to Cityscapes | mIoU: 30.89 -> 32.29 | Mixed |
| patch size 8 instead of 16 for feature extraction (ablation) | VSAIT · GTA to Cityscapes | mIoU: 30.89 -> 31.47 | No clear effect |
| VSAIT versus best baseline SRUNIT on segmentation | VSAIT · GTA to Cityscapes | mIoU: 22.69 -> 30.89 | Improved |
| VSAIT versus EPE on KID | VSAIT · GTA to Cityscapes | KID: 10.95 -> 8.74 | Improved |
| VSAIT versus SRUNIT on pixel accuracy delta1 in photo to map | VSAIT · Aerial Photo to Google Map | pixel accuracy delta1: 17.67 -> 15.2 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DRIT — DRIT++: Diverse Image-to-Image Translation via Disentangled Representations (IJCV 2020) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Unpaired Image-To-Image Translation Using Cycle-Consistent Adversarial Networks (ICCV 2017 · similarity 0.89)
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.88)
- Unsupervised Image-to-Image Translation Using Domain-Specific Variational Information Bound (NeurIPS 2018 · similarity 0.88)
- Theoretical Analysis of Image-to-Image Translation with Adversarial Learning (ICML 2018 · similarity 0.88)
- Semantically Robust Unpaired Image Translation for Data With Unmatched Semantics Statistics (ICCV 2021 · similarity 0.88)
- Few-Shot Unsupervised Image-to-Image Translation (ICCV 2019 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
