Research map / Detection and segmentation / CNN features
RevColV2: Exploring Disentangled Representations in Masked Image Modeling
Qi Han, Yuxuan Cai, Xiangyu Zhang
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 |
|---|---|---|---|
| keep top-down decoder columns during fine-tuning (vs encoder only) (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 83.8 -> 84.7 | Improved |
| keep decoder, matching params/FLOPs for encoder-only variant (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 83.9 -> 84.7 | Improved |
| keep vanilla MAE decoder in fine-tuning (ablation) | ViT + MAE · ImageNet-1K | top-1 accuracy: 83.6 -> 83.2 | Hurt |
| randomly initialize decoder, use only pre-trained encoder (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 84.7 -> 84.4 | Hurt |
| randomly initialize decoder, use only pre-trained encoder (ablation) | RevColV2-B · ADE20K | mIoU: -0.6 | Hurt |
| increase decoder column depth from 4 to 8 (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 84.1 -> 83.6 | Hurt |
| increase decoder column depth from 4 to 12 (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 84.1 -> 82.9 | Hurt |
| increase decoder column depth from 4 to 8 (ablation) | RevColV2-L · ImageNet-1K | top-1 accuracy: 85.2 -> 84.9 | Hurt |
| pre-train 1600 epochs instead of 300 (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 84.1 -> 84.7 | Improved |
| pre-train 1600 epochs instead of 300 (ablation) | RevColV2-L · ImageNet-1K | top-1 accuracy: 85.2 -> 86.3 | Improved |
| replace reversible column decoder with MAE ViT-block decoder (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 84.1 -> 83.5 | Hurt |
| replace reversible column decoder with UNet-like single column (ablation) | RevColV2-B · ImageNet-1K | top-1 accuracy: 84.1 -> 83.4 | Hurt |
Closest papers by meaning
- Reversible Column Networks (ICLR 2023 · similarity 0.88)
- MIM-Refiner: A Contrastive Learning Boost from Intermediate Pre-Trained Masked Image Modeling Representations (ICLR 2025 · similarity 0.88)
- MaskAnyNet: Rethinking Masked Image Regions as Valuable Information in Supervised Learning (AAAI 2026 · similarity 0.86)
- Towards Latent Masked Image Modeling for Self-Supervised Visual Representation Learning (ECCV 2024 · similarity 0.86)
- Stare at What You See: Masked Image Modeling Without Reconstruction (CVPR 2023 · similarity 0.86)
- MaskViM: Domain Generalized Semantic Segmentation with State Space Models (AAAI 2025 · similarity 0.86)
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.
