Research map / Detection and segmentation / Open-set and few-shot detection
Vector-Decomposed Disentanglement for Domain-Invariant Object Detection
Aming Wu, Rui Liu, Yahong Han, Linchao Zhu, Yi Yang
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 |
|---|---|---|---|
| plug in vector-decomposed disentanglement | SW · Cityscapes to FoggyCityscapes | mAP (%): 34.3 -> 37.9 | Improved |
| plug in vector-decomposed disentanglement | ICCR · Cityscapes to FoggyCityscapes | mAP (%): 37.4 -> 40.0 | Improved |
| plug in vector-decomposed disentanglement | SW · Daytime-sunny to Dusk-rainy | mAP (%): 31.5 -> 36.9 | Improved |
| plug in vector-decomposed disentanglement | ICCR · Daytime-sunny to Dusk-rainy | mAP (%): 34.9 -> 37.8 | Improved |
| plug in vector-decomposed disentanglement | SW · Daytime-sunny to Night-rainy | mAP (%): 17.4 -> 23.1 | Improved |
| plug in vector-decomposed disentanglement | ICCR · Daytime-sunny to Night-rainy | mAP (%): 20.6 -> 23.7 | Improved |
| plug in vector-decomposed disentanglement | SW · Daytime-sunny to compound target (Dusk+Night rainy) | mAP (%): 27.3 -> 32.6 | Improved |
| use two-step training instead of one-step training (ablation) | SW-VDD · Cityscapes to FoggyCityscapes | mAP (%): 34.5 -> 37.9 | Improved |
| use two-step training instead of one-step training (ablation) | SW-VDD · PASCAL to Watercolor | mAP (%): 54.5 -> 56.6 | Improved |
| add orthogonal loss with two-step training (ablation) | SW-VDD · Cityscapes to FoggyCityscapes | mAP (%): 36.5 -> 37.9 | Improved |
| add orthogonal loss with two-step training (ablation) | SW-VDD · PASCAL to Watercolor | mAP (%): 54.9 -> 56.6 | Improved |
| add orthogonal loss with one-step training (ablation) | SW-VDD · Cityscapes to FoggyCityscapes | mAP (%): 33.2 -> 34.5 | Improved |
Closest papers by meaning
- Triple Feature Disentanglement for One-Stage Adaptive Object Detection (AAAI 2024 · similarity 0.94)
- DA-Ada: Learning Domain-Aware Adapter for Domain Adaptive Object Detection (NeurIPS 2024 · similarity 0.89)
- SEEN-DA: SEmantic ENtropy guided Domain-aware Attention for Domain Adaptive Object Detection (CVPR 2025 · similarity 0.89)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.88)
- Non-exemplar Domain Incremental Object Detection via Learning Domain Bias (AAAI 2024 · similarity 0.88)
- DSD-DA: Distillation-based Source Debiasing for Domain Adaptive Object Detection (ICML 2024 · similarity 0.88)
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.
