Research map / Detection and segmentation / CNN features
Quantum-soft QUBO Suppression for Accurate Object Detection
Junde Li, Swaroop Ghosh
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 |
|---|---|---|---|
| add soft-NMS rescoring to basic QUBO suppression (ablation) | QQS · MS-COCO test-dev | AP 0.5:0.95: 21.8 -> 25.7 | Improved |
| add penalty and reward adjustment terms (ablation) | QSQS · MS-COCO test-dev | AP 0.5:0.95: 25.7 -> 25.8 | No clear effect |
| replace greedy NMS with QSQS | Faster R-CNN · MS-COCO test-dev | AP 0.5:0.95: 25.6 -> 25.7 | Improved |
| replace greedy NMS with QSQS | Faster R-CNN · MS-COCO test-dev | AR@100: 38.4 -> 39.4 | Improved |
| replace greedy NMS with basic QUBO (QQS) | Faster R-CNN · MS-COCO test-dev | AP@0.5: 43.9 -> 35.4 | Hurt |
| replace greedy NMS with QSQS-enh | Faster R-CNN · PASCAL VOC 2007 | mAP (%): 74.20 -> 75.11 | Improved |
| replace greedy NMS with QSQS | Faster R-CNN · PASCAL VOC 2007 | inference latency (ms): 65.87 -> 138.9 | Hurt |
| replace NMS with QSQS | CSP · CityPersons Reasonable | MR: 27.48 -> 27.33 | Improved |
| replace NMS with QSQS | CSP · CityPersons Heavy | MR: 65.78 -> 66.38 | Hurt |
| replace NMS with soft-NMS | CSP · CityPersons Reasonable | MR: 27.48 -> 28.13 | Hurt |
| replace NMS with QSQS | CSP · CityPersons Partial | MR: 27.34 -> 27.48 | Hurt |
| use overlap threshold 0.3 or 0.7 instead of 0.5 (ablation) | QSQS · PASCAL VOC 2007 | mAP (%) | Hurt |
Closest papers by meaning
- Soft-NMS -- Improving Object Detection With One Line of Code (ICCV 2017 · similarity 0.85)
- Accelerating Non-Maximum Suppression: A Graph Theory Perspective (NeurIPS 2024 · similarity 0.85)
- Q-DETR: An Efficient Low-Bit Quantized Detection Transformer (CVPR 2023 · similarity 0.83)
- Reg-PTQ: Regression-specialized Post-training Quantization for Fully Quantized Object Detector (CVPR 2024 · similarity 0.83)
- Learning Non-Maximum Suppression (CVPR 2017 · similarity 0.83)
- AQ-DETR: Low-Bit Quantized Detection Transformer with Auxiliary Queries (AAAI 2024 · similarity 0.81)
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.
