Research map / 3D vision / 3D generation
4Seasons: Benchmarking Visual SLAM and Long-Term Localization for Autonomous Driving in Challenging Conditions
Patrick Wenzel, Nan Yang, Rui Wang, Niclas Zeller, Daniel Cremers
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 IMU to stereo input | Basalt · 4Seasons VO, unknown scenarios | % high precision: 14.0 -> 21.9 | Improved |
| add IMU to stereo input | ORB-SLAM3 · 4Seasons VO, unknown scenarios | % high precision: 0.2 -> 15.0 | Improved |
| add IMU to stereo input | Basalt · 4Seasons VO, known scenarios | % medium precision: 32.3 -> 30.3 | No clear effect |
| add IMU to stereo input | ORB-SLAM3 · 4Seasons VO, known scenarios | % high precision: 3.9 -> 9.4 | Mixed |
| evaluate at night instead of afternoon | Basalt (stereo) · Maximilianeum VO | % high / medium / coarse precision: 35.1 / 75.3 / 91.8 -> 1.2 / 9.8 / 38.2 | Hurt |
| evaluate in cloudy instead of sunny weather | ORB-SLAM3 (stereo) · Office Loop VO | % high / medium / coarse precision: 16.8 / 65.3 / 94.9 -> 1.4 / 24.0 / 82.2 | Hurt |
| use SuperPoint+SuperGlue instead of SIFT+NN | hloc · 4Seasons MBVL average | % within 0.1 m/1 deg: 11.5 -> 40.3 | Improved |
| use R2D2+NN instead of SIFT+NN | hloc · 4Seasons MBVL average | % within 1 m/5 deg: 30.2 -> 75.9 | Improved |
| use NetVLAD instead of CIR or DIR | global place recognition · 4Seasons GPR average | % within 1 m/5 deg: 22.2 -> 34.8 | Improved |
| use SuperPoint+SuperGlue instead of D2-Net+NN | hloc · 4Seasons MBVL average | % within 0.1 m/1 deg: 21.5 -> 40.3 | Improved |
| evaluate on unknown instead of known scenarios | ORB-SLAM3 (stereo) · 4Seasons VO average | % medium precision: 31.3 -> 5.9 | Hurt |
Closest papers by meaning
- Benchmarking 6DOF Outdoor Visual Localization in Changing Conditions (CVPR 2018 · similarity 0.88)
- CrowdDriven: A New Challenging Dataset for Outdoor Visual Localization (ICCV 2021 · similarity 0.87)
- Benchmarking Egocentric Visual-Inertial SLAM at City Scale (ICCV 2025 · similarity 0.87)
- Cam4DOcc: Benchmark for Camera-Only 4D Occupancy Forecasting in Autonomous Driving Applications (CVPR 2024 · similarity 0.86)
- Beyond Chamfer Distance: Granular Order-aware Evaluation Metric For Online Mapping (NeurIPS 2026 · similarity 0.86)
- ZeroVO: Visual Odometry with Minimal Assumptions (CVPR 2025 · 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.
