Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly Supervised Knowledge Transfer with Probabilistic Logical Reasoning for Object Detection
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 |
|---|---|---|---|
| add ProbKT probabilistic logical supervision (class counts) on top of pre-trained RCNN | RCNN pre-trained on source domain · CLEVR-mini target domain | count accuracy: 0.0 -> 0.975 | Helped |
| use ProbKT (class-count logical supervision) instead of WSOD-transfer | RCNN backbone · Molecules target domain | count accuracy: 0.001 -> 0.942 | Helped |
| add ProbKT with RCNN backbone | RCNN pre-trained · MNIST target domain | sum accuracy: 0.298 -> 0.903 | Helped |
| switch backbone from RCNN to DETR in ProbKT (ablation) | ProbKT · Molecules target domain | count accuracy: 0.942 -> 0.508 | Hurt |
| use most-probable-world (Hungarian) approximation ProbKT* instead of full ProbKT (ablation) | ProbKT with RCNN · CLEVR-mini target domain | count accuracy: 0.975 vs 0.974 | No clear effect |
| use most-probable-world (Hungarian) approximation ProbKT* instead of full ProbKT (ablation) | ProbKT with RCNN · Molecules target domain | count accuracy: 0.942 -> 0.945 (ProbKT* slightly higher on target but lower OOD 0.592 vs 0.603) | Hurt |
| iterative relabeling after fine-tuning (ablation) | ProbKT · CLEVR, Molecules, MNIST | performance on test sets | Helped |
| iterative relabeling after fine-tuning (ablation) | ProbKT with DETR · Molecules | performance on test sets | No clear effect |
Closest papers by meaning
- W2F: A Weakly-Supervised to Fully-Supervised Framework for Object Detection (CVPR 2018 · similarity 0.89)
- Revisiting Knowledge Transfer for Training Object Class Detectors (CVPR 2018 · similarity 0.89)
- Weak-Shot Object Detection Through Mutual Knowledge Transfer (CVPR 2023 · similarity 0.89)
- Mixed Supervised Object Detection by Transferring Mask Prior and Semantic Similarity (NeurIPS 2021 · similarity 0.89)
- UniT: Unified Knowledge Transfer for Any-Shot Object Detection and Segmentation (CVPR 2021 · similarity 0.88)
- Relational Matching for Weakly Semi-Supervised Oriented Object Detection (CVPR 2024 · similarity 0.86)
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.