Research map / Architectures and efficiency / RNNs and normalization
DSAC - Differentiable RANSAC for Camera Localization
Check 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 |
|---|---|---|---|
| Train DSAC end-to-end rather than componentwise (ablation) | DSAC camera localization pipeline · 7-Scenes Complete | accuracy: 60.3% -> 62.5%; +2.2% | Helped |
| Train SoftAM end-to-end rather than componentwise (ablation) | SoftAM camera localization pipeline · 7-Scenes Complete | accuracy: 61.6% -> 57.8%; -3.8% | Hurt |
| Use end-to-end DSAC instead of Brachmann et al.'s pipeline | 7-Scenes camera localization · 7-Scenes Complete | accuracy: 55.2% -> 62.5%; 7.3% | Helped |
| Use probabilistic hypothesis selection instead of soft argmax selection (ablation) | End-to-end camera localization pipeline · 7-Scenes Complete | accuracy: 57.8% -> 62.5% | Helped |
| Use probabilistic hypothesis selection instead of deterministic RANSAC selection (ablation) | Componentwise-trained camera localization pipeline · 7-Scenes Complete | accuracy: 61.0% -> 60.3%; -0.7% | Hurt |
| Train both the Coordinate CNN and Score CNN end-to-end (ablation) | DSAC camera localization pipeline · 7-Scenes Complete | accuracy | Helped |
| Use soft argmax, which averages candidate hypotheses, instead of hard probabilistic selection (ablation) | Differentiable RANSAC camera localization · 7-Scenes | accuracy: 62.5% -> 57.8% (Complete, end-to-end) | Hurt |
| Replace soft argmax selection with probabilistic selection after end-to-end training (ablation) | Differentiable RANSAC camera localization · 7-Scenes Complete | accuracy: 57.2% -> 62.4% | Helped |
Closest papers by meaning
- RANSAC Scoring Done Right (NeurIPS 2026 · similarity 0.81)
- D-FINE: Redefine Regression Task of DETRs as Fine-grained Distribution Refinement (ICLR 2025 · similarity 0.79)
- SDAC: A Multimodal Synthetic Dataset for Anomaly and Corner Case Detection in Autonomous Driving (AAAI 2024 · similarity 0.79)
- Reinforcement Learning as a Parsimonious Alternative to Prediction Cascades: A Case Study on Image Segmentation (AAAI 2024 · similarity 0.79)
- 2DMamba: Efficient State Space Model for Image Representation with Applications on Giga-Pixel Whole Slide Image Classification (CVPR 2025 · similarity 0.78)
- Deep Homography Estimation for Visual Place Recognition (AAAI 2024 · similarity 0.78)
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.