Research map / Detection and segmentation / CNN features
1% VS 100%: Parameter-Efficient Low Rank Adapter for Dense Predictions
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 LoRand adapters instead of full fine-tuning | Swin-L · PASCAL VOC 0712 val | APBox: 84.43 % -> 87.12 % | Helped |
| Use LoRand adapters instead of full fine-tuning | Swin-L · ADE20K val | mIoU: 53.25 % -> 50.67 % | Hurt |
| Use LoRand adapters instead of full fine-tuning | Swin-B · COCO 2017 val | APBox: 51.90 % -> 51.10 % | Hurt |
| Use LoRand adapters instead of full fine-tuning | Swin-B · COCO 2017 val | APMask: 45.00 % -> 44.10 % | Hurt |
| Increase branch number α from 2 to 4 (ablation) | LoRand on Swin-L · PASCAL VOC 0712 val | APBox: 87.12 % -> 87.63 % | Helped |
| Increase branch number α from 2 to 4 (ablation) | LoRand on Swin-L · ADE20K val | mIoU: 50.67 % -> 51.13 % | Helped |
| Increase branch number α from 2 to 4 (ablation) | LoRand on Swin-B · COCO 2017 val | APBox: 51.10 % -> 51.20 % | Helped |
| Increase branch number α from 2 to 4 (ablation) | LoRand on Swin-B · COCO 2017 val | APMask: 44.10 % -> 44.30 % | Helped |
| Increase kernel dimension β from 8 to 16, with α=4 (ablation) | LoRand on Swin-L · PASCAL VOC 0712 val | APBox: 87.63 % -> 88.11 % | Helped |
| Increase kernel dimension β from 8 to 16, with α=4 (ablation) | LoRand on Swin-L · ADE20K val | mIoU: 51.13 % -> 51.87 % | Helped |
| Increase kernel dimension β from 8 to 16, with α=4 (ablation) | LoRand on Swin-B · COCO 2017 val | APBox: 51.20 % -> 51.50 % | Helped |
| Increase kernel dimension β from 8 to 16, with α=4 (ablation) | LoRand on Swin-B · COCO 2017 val | APMask: 44.30 % -> 44.40 % | Helped |
Closest papers by meaning
- PELA: Learning Parameter-Efficient Models with Low-Rank Approximation (CVPR 2024 · similarity 0.88)
- C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models (NeurIPS 2026 · similarity 0.87)
- RandLoRA: Full rank parameter-efficient fine-tuning of large models (ICLR 2025 · similarity 0.86)
- Minimal Interaction Seperated Tuning: A New Paradigm for Visual Adaptation (CVPR 2025 · similarity 0.86)
- VFM-Adapter: Adapting Visual Foundation Models for Dense Prediction with Dynamic Hybrid Operation Mapping (AAAI 2025 · similarity 0.86)
- LoRAtorio: An intrinsic approach to LoRA Skill Composition (NeurIPS 2026 · 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.