Research map / Detection and segmentation / Open-set and few-shot detection
Revisiting Computer-Aided Tuberculosis Diagnosis
Yun Liu, Yu-Huan Wu, Shichen Zhang, Li Liu, Min Wu, Ming‐Ming Cheng
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 SymFormer to RetinaNet | RetinaNet · TBX11K test, all CXR images | AP50 bb (%): 52.1 -> 68.0 | Improved |
| add SymFormer to Deformable DETR | Deformable DETR · TBX11K test, all CXR images | AP50 bb (%): 51.7 -> 57.0 | Improved |
| add SymFormer to RetinaNet | RetinaNet · TBX11K test (CXR classification) | accuracy (%): 87.4 -> 94.5 | Improved |
| add SymFormer to Deformable DETR | Deformable DETR · DA+DB test (no fine-tuning) | accuracy (%): 68.6 -> 82.4 | Improved |
| add SymFormer to RetinaNet | RetinaNet · MC+Shenzhen test (no fine-tuning) | accuracy (%): 49.3 -> 82.8 | Improved |
| use RetinaNet instead of Deformable DETR as the SymFormer base | SymFormer · TBX11K test, all CXR images | AP50 bb (%): 57.0 -> 68.0 | Improved |
| use P2T-Small backbone instead of ResNet-50 | SymFormer w/ RetinaNet · TBX11K test (CXR classification) | accuracy (%): 94.5 -> 94.6; FPS 24.3 -> 17.9 | No clear effect |
| add symmetric search attention to SPE (vs vanilla attention) (ablation) | SymFormer w/ RetinaNet · TBX11K val, TB images | AP50 bb (%): 75.7 -> 76.6 | Improved |
| add symmetric positional encoding instead of APE under SymAttention (ablation) | SymFormer w/ RetinaNet · TBX11K val, TB images | AP50 bb (%): 74.9 -> 76.6 | Improved |
| add STN to SPE (ablation) | SymFormer w/ RetinaNet · TBX11K val, TB images | AP50 bb (%): 75.5 -> 76.6 | Improved |
| transfer positional encoding right to left instead of left to right (ablation) | SymFormer w/ RetinaNet · TBX11K val, TB images | AP50 bb (%): 76.3 -> 76.6 | Improved |
| add vanilla attention with RPE (ablation) | Detector without attention · TBX11K val, TB images | AP50 bb (%): 72.7 -> 72.7 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- Deformable DETR — Deformable DETR: Deformable Transformers for End-to-End Object Detection (ICLR 2021) · better on 6 of 6 shared results
- FCOS — FCOS: Fully Convolutional One-Stage Object Detection (ICCV 2019) · better on 1 of 1 shared result
- Faster R-CNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · better on 1 of 1 shared result
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
- Novel Class Discovery in Chest X-rays via Paired Images and Text (AAAI 2024 · similarity 0.83)
- CXReasonBench: A Benchmark for Evaluating Structured Diagnostic Reasoning in Chest X-rays (NeurIPS 2025 · similarity 0.82)
- CARZero: Cross-Attention Alignment for Radiology Zero-Shot Classification (CVPR 2024 · similarity 0.81)
- BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays (NeurIPS 2024 · similarity 0.80)
- RadZero: Similarity-Based Cross-Attention for Explainable Vision-Language Alignment in Chest X-ray with Zero-Shot Multi-Task Capability (NeurIPS 2025 · similarity 0.80)
- Learning Low-Rank Feature for Thorax Disease Classification (NeurIPS 2024 · similarity 0.80)
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.
