Research map / Vision-language and multimodal / CLIP and zero-shot
Mix and Match Networks: Encoder-Decoder Alignment for Zero-Pair Image Translation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use direct zero-pair depth-to-segmentation translation instead of translating through RGB | M&MNet · SceneNetRGBD 170K, depth-to-segmentation | mIoU: 24.7 -> 55.4 | Helped |
| Fuse RGB with depth as input to segmentation | M&MNet D→S · SceneNetRGBD 170K, depth-to-segmentation | mIoU: 55.4 -> 57.1 | Helped |
| Add pooling indices as side information (ablation) | M&MNet · SceneNetRGBD 51K, depth-to-segmentation | mIoU: 32.2% -> 45.6% | Helped |
| Replace pooling indices with skip connections (ablation) | M&MNet · SceneNetRGBD 51K, depth-to-segmentation | mIoU: 32.2% -> 14.1% | Hurt |
| Initialize the RGB encoder with ImageNet-pretrained weights (ablation) | M&MNet · SceneNetRGBD 51K, depth-to-segmentation | mIoU: 45.6% -> 49.5% | Helped |
| Add autoencoders (ablation) | M&MNet · SceneNetRGBD 51K, depth-to-segmentation | mIoU: 5.64% -> 22.9% | Helped |
| Add cross-domain latent-space consistency (ablation) | M&MNet with autoencoders · SceneNetRGBD 51K, depth-to-segmentation | mIoU: 22.9% -> 48.9% | Helped |
| Add noise to the encoder latent output (ablation) | M&MNet with autoencoders and latent consistency · SceneNetRGBD 51K, depth-to-segmentation | mIoU: 48.9% -> 49.5% | Helped |
Closest papers by meaning
- Do Vision and Language Encoders Represent the World Similarly? (CVPR 2024 · similarity 0.87)
- Cross-Domain Correspondence Learning for Exemplar-Based Image Translation (CVPR 2020 · similarity 0.86)
- DH-Set: Improving Vision-Language Alignment with Diverse and Hybrid Set-Embeddings Learning (CVPR 2025 · similarity 0.85)
- Harnessing Frozen Unimodal Encoders for Flexible Multimodal Alignment (CVPR 2025 · similarity 0.85)
- SynthRGB-T: Language-Vision Guided Image Translation for Diversity Synthesis (CVPR 2026 · similarity 0.85)
- Rethinking the Paradigm of Content Constraints in Unpaired Image-to-Image Translation (AAAI 2024 · similarity 0.85)
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.