Research map / Detection and segmentation / Open-set and few-shot detection
Expanding Low-Density Latent Regions for Open-Set Object Detection
Check 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 |
|---|---|---|---|
| Add CFL and UPL together (ablation) | Faster R-CNN OSOD baseline · VOC-COCO-20 | WI: 19.26 -> 14.95 | Helped |
| Add CFL and UPL together (ablation) | Faster R-CNN OSOD baseline · VOC-COCO-20 | AP_U: 0 -> 14.93 | Helped |
| Add CFL alone (ablation) | Faster R-CNN OSOD baseline · VOC-COCO-20 | WI: 19.26 -> 17.92 | Helped |
| Add UPL alone (ablation) | Faster R-CNN OSOD baseline · VOC-COCO-20 | AP_U: 0 -> 14.27 | Helped |
| Add hard example mining (ablation) | OpenDet UPL · VOC-COCO-20 | WI: 18.33 -> 14.95 | Helped |
| Include background proposals in mining, rather than sampling foreground only (ablation) | OpenDet UPL · VOC-COCO-20 | mAP_K: 56.53 -> 58.75 | Helped |
| Use uncertainty weight (1-p_{c*})^α p_{c*} instead of identity weighting (ablation) | OpenDet UPL · VOC-COCO-20 | mAP_K: 56.42 -> 58.75 | Helped |
| Use class-balanced memory rather than class-agnostic memory of size 5120 (ablation) | OpenDet CFL · VOC-COCO-20 | mAP_K: 57.47 -> 58.75 | Helped |
| Replace end-to-end training with UPL fine-tuning (ablation) | OpenDet · VOC-COCO-20 | WI: 14.95 -> 17.01 | Hurt |
| Increase CFL loss weight γ_t to 1.0 (ablation) | OpenDet · VOC-COCO-20 | mAP_K: 58.75 -> 53.36 | Hurt |
| Use a higher score threshold for testing | Faster R-CNN baseline · VOC-COCO-20 | WI: 18.39 -> 18.83 | No clear effect |
| Increase the number of dropout-sampling runs (ablation) | DS · VOC-COCO-20 | WI: 15.26 -> 17.01 | Hurt |
Closest papers by meaning
- Towards Evidential and Class Separable Open Set Object Detection (AAAI 2024 · similarity 0.88)
- Open-World Objectness Modeling Unifies Novel Object Detection (CVPR 2025 · similarity 0.87)
- VL-SAM-V2: Open-World Object Detection with General and Specific Query Fusion (NeurIPS 2025 · similarity 0.87)
- Percept, Memory, and Imagine: World Feature Simulating for Open-Domain Unknown Object Detection (CVPR 2025 · similarity 0.87)
- Training-Free Open-Ended Object Detection and Segmentation via Attention as Prompts (NeurIPS 2024 · similarity 0.87)
- Generative Region-Language Pretraining for Open-Ended Object Detection (CVPR 2024 · similarity 0.87)
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.