Research map / Detection and segmentation / Weakly supervised detection and segmentation
STEP-DETR: Advancing DETR-based Semi-Supervised Object Detection with Super Teacher and Pseudo-Label Guided Text Queries
Tahira Shehzadi, Khurram Azeem Hashmi, Shalini Sarode, Didier Stricker, Muhammad Zeshan Afzal
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 Pseudo-Label Text Queries (ablation) | Semi-DETR · COCO 10% labeled | mAP: 43.5 -> 44.7 | Improved |
| add Denoising Text Guided Object Queries (ablation) | Semi-DETR + text queries · COCO 10% labeled | mAP: 44.7 -> 45.1 | Improved |
| add Query Refinement (ablation) | Semi-DETR + text and denoising queries · COCO 10% labeled | mAP: 45.1 -> 45.4 | Improved |
| use text queries instead of sparse visual queries (ablation) | STEP-DETR · COCO 10% labeled | mAP: 44.3 -> 45.4 | Improved |
| use consistency queries instead of standard queries (ablation) | DINO teacher-student SSOD · COCO 10% labeled | mAP: 41.3 -> 43.5 | Improved |
| add Super Teacher (ablation) | DINO · COCO 10% labeled | mAP: 35.0 -> 45.4 | Improved |
| add NMS on top of Super Teacher (ablation) | STEP-DETR · COCO 10% labeled | mAP: 45.4 -> 45.7 | No clear effect |
| use text guided denoising instead of standard denoising (ablation) | Semi-DETR · COCO 10% labeled | mAP: 43.5 -> 43.8 | Improved |
| use query similarity refinement instead of simple concat (ablation) | STEP-DETR · COCO 10% labeled | mAP: 45.1 -> 45.4 | Improved |
| add Query Refinement to reduce redundant queries | STEP-DETR · COCO one-to-many training | training time per 1k iterations: 38.56 -> 30.98 | Improved |
| STEP-DETR versus Semi-DETR with 10% labels | STEP-DETR · COCO-Partial 10% labeled | mAP: 43.5 -> 45.4 | Improved |
| STEP-DETR versus Semi-DETR on full COCO plus unlabeled | STEP-DETR · COCO-Full | mAP: 50.4 -> 52.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Sparse Semi-DETR — Sparse Semi-DETR: Sparse Learnable Queries for Semi-Supervised Object Detection (CVPR 2024) · better on 5 of 5 shared results
- Omni-DETR — Omni-DETR: Omni-Supervised Object Detection With Transformers (CVPR 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Sparse Semi-DETR: Sparse Learnable Queries for Semi-Supervised Object Detection (CVPR 2024 · similarity 0.93)
- Toward Efficient Semi-Supervised Object Detection With Detection Transformer (TPAMI 2025 · similarity 0.93)
- Semi-DETR: Semi-Supervised Object Detection With Detection Transformers (CVPR 2023 · similarity 0.93)
- Diverse Learner: Exploring Diverse Supervision for Semi-Supervised Object Detection (ECCV 2022 · similarity 0.90)
- Humble Teachers Teach Better Students for Semi-Supervised Object Detection (CVPR 2021 · similarity 0.89)
- Unbiased Teacher for Semi-Supervised Object Detection (ICLR 2021 · 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.
