Research map / Detection and segmentation / Open-set and few-shot detection
Seeking Similarities Over Differences: Similarity-Based Domain Alignment for Adaptive Object Detection
Farzaneh Rezaeianaran, Rakshith Shetty, Rahaf Aljundi, Daniel Olmeda Reino, Shanshan Zhang, Bernt Schiele
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 |
|---|---|---|---|
| use adversarial training instead of contrastive loss (single group) (ablation) | Faster R-CNN UDA · Sim2Real | car AP: 33.2 -> 40.8 | Improved |
| use adversarial training instead of contrastive loss (multiple groups) (ablation) | Faster R-CNN UDA · Sim2Real | car AP: 36.9 -> 43.1 | Improved |
| use adversarial training instead of contrastive loss (MG+CA) (ablation) | Faster R-CNN UDA · Sim2Real | car AP: 42.6 -> 45.6 | Improved |
| add image-level alignment (multiple groups, class-agnostic) (ablation) | ViSGA · Sim2Real | car AP: 45.6 -> 49.3 | Improved |
| add image-level alignment (single group) (ablation) | ViSGA single group · Sim2Real | car AP: 40.8 -> 39.5 | Hurt |
| use multiple visually similar groups instead of no grouping (ablation) | ViSGA · Sim2Real | car AP: 39.0 -> 44.9 | Improved |
| use multiple visually similar groups instead of no grouping (ablation) | ViSGA · Foggy | mAP: 38.5 -> 41.8 | Improved |
| aggregate into a single group instead of no grouping (ablation) | ViSGA · Foggy | mAP: 38.5 -> 33.7 | Hurt |
| ignore predicted class label in grouping (class-agnostic) (ablation) | ViSGA · Sim2Real | car AP: 44.9 -> 49.3 | Improved |
| ignore predicted class label in grouping (class-agnostic) (ablation) | ViSGA · Foggy | mAP: 41.8 -> 43.3 | Improved |
| use adaptive number of clusters instead of fixed (ablation) | ViSGA · Foggy | mAP: 42.5 -> 43.3 | Improved |
| use cosine similarity instead of IoU for grouping (ablation) | ViSGA · Sim2Real | car AP: 44.8 -> 49.3 | Improved |
Closest papers by meaning
- Adapting Object Detectors via Selective Cross-Domain Alignment (CVPR 2019 · similarity 0.93)
- Strong-Weak Distribution Alignment for Adaptive Object Detection (CVPR 2019 · similarity 0.92)
- iFAN: Image-Instance Full Alignment Networks for Adaptive Object Detection (AAAI 2020 · similarity 0.91)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · similarity 0.91)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.90)
- Differential Alignment for Domain Adaptive Object Detection (AAAI 2025 · similarity 0.89)
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.
