Research map / Detection and segmentation / CNN features
Accurate Detection of Proteins in Cryo-Electron Tomograms from Sparse Labels
Qinwen Huang, Ye Zhou, Hsuan-Fu Liu, Alberto Bartesaghi
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 voxel-level debiased contrastive learning module (ablation) | PU detector · EMPIAR-10304 (70% labels) | mAP r5: 65.8 -> 72.1 | Improved |
| add voxel-level debiased contrastive learning module (ablation) | PU detector · EMPIAR-10499 (5% labels) | mAP r5: 24.2 -> 31.1 | Improved |
| add voxel-level debiased contrastive learning module (ablation) | PU detector · EMPIAR-10499 (5% labels) | mAP r2: 14.1 -> 22.0 | Improved |
| add positive-unlabeled learning in both modules instead of treating unlabeled as negative (ablation) | detector · EMPIAR-10304 (5% labels) | mAP r5: 32.2 -> 54.5 | Improved |
| add positive-unlabeled learning in both modules (ablation) | detector · EMPIAR-10499 (70% labels) | mAP r5: 20.0 -> 42.5 | Improved |
| add positive-unlabeled learning in both modules (ablation) | detector · EMPIAR-10499 (5% labels) | mAP r2: 2.1 -> 22.0 | Improved |
| increase annotated fraction from 5% to 30% | Ours · EMPIAR-10304 | F1: 68.2 -> 78.0 | Improved |
| increase annotated fraction from 5% to 70% | Ours · EMPIAR-10499 | F1: 53.5 -> 60.2 | Improved |
| ours (5% labels) vs template matching | Ours · EMPIAR-10499 | F1: 35.5 -> 53.5 | Improved |
| ours (70% labels) vs crYOLO | Ours · EMPIAR-10304 | F1: 22.4 -> 80.0 | Improved |
| use 2D-then-3D convolution backbone instead of full 3D ResNet/UNet (ablation) | detector · CET datasets | detection | Improved |
| increase network depth (ablation) | detector backbone · CET datasets | detection | Hurt |
Closest papers by meaning
- At FullTilt: Real-Time Open-Set 3D Macromolecule Detection Directly from Tilted 2D Projections (NeurIPS 2026 · similarity 0.88)
- cryoSPHERE: Single-Particle HEterogeneous REconstruction from cryo EM (ICLR 2025 · similarity 0.87)
- Vox-UDA: Voxel-wise Unsupervised Domain Adaptation for Cryo-Electron Subtomogram Segmentation with Denoised Pseudo-Labeling (AAAI 2025 · similarity 0.86)
- BOE-ViT: Boosting Orientation Estimation with Equivariance in Self-Supervised 3D Subtomogram Alignment (CVPR 2025 · similarity 0.86)
- CryoFM: A Flow-based Foundation Model for Cryo-EM Densities (ICLR 2025 · similarity 0.85)
- Reconstructing Heterogeneous Biomolecules via Hierarchical Gaussian Mixtures and Part Discovery (NeurIPS 2025 · similarity 0.85)
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.
