Research map / Label-efficient and robust learning / Self-supervised and contrastive
ActiveZero: Mixed Domain Learning for Active Stereovision With Zero Annotation
Isabella Liu, Edward Yang, Jianyu Tao, Rui Chen, Xiaoshuai Zhang, Qing Ran, Zhu Liu, Hao Su
Check 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 |
|---|---|---|---|
| Use patch size 15 instead of patch size 11 (ablation) | ActiveZero patch-wise reprojection loss · Testing dataset | Abs depth err (mm): 5.115 -> 5.114 | Improved |
| Use patch size 21 instead of patch size 11 (ablation) | ActiveZero patch-wise reprojection loss · Testing dataset | > 4mm: 0.393 -> 0.385 | Improved |
| Use patch size 7 instead of patch size 11 (ablation) | ActiveZero patch-wise reprojection loss · Testing dataset | Abs depth err (mm): 5.115 -> 5.507 | Hurt |
| Set λs=1 and λr=0.5 instead of λs=0.01 and λr=2 (ablation) | ActiveZero · Testing dataset | Abs depth err (mm): 4.377 -> 7.578 | Hurt |
| Set λs=1 and λr=1 instead of λs=0.01 and λr=2 (ablation) | ActiveZero · Testing dataset | Abs depth err (mm): 4.377 -> 6.064 | Hurt |
| Set λs=1 and λr=2 instead of λs=0.01 and λr=2 (ablation) | ActiveZero · Testing dataset | Abs depth err (mm): 4.377 -> 5.672 | Hurt |
| Set λs=0.05 and λr=2 instead of λs=0.01 and λr=2 (ablation) | ActiveZero · Testing dataset | Abs depth err (mm): 4.377 -> 5.543 | Hurt |
| Set λs=0.002 and λr=2 instead of λs=0.01 and λr=2 (ablation) | ActiveZero · Testing dataset | Abs depth err (mm): 4.377 -> 4.683 | Hurt |
| Add the 6-layer filter (ablation) | ActiveZero · Testing dataset | Abs depth err (mm): 4.592 -> 4.377 | Improved |
| Add the 6-layer filter (ablation) | ActiveZero · Testing dataset | > 4mm: 0.356 -> 0.335 | Improved |
| Use simulation disparity loss (ablation) | ActiveZero · Simulation training | Convergence rate | Improved |
| Use ActiveZero instead of StereoGAN+PSM | Stereo inference · Inference, 960×540 image pair | Inference Time(s): 0.303 -> 0.256 | Improved |
Closest papers by meaning
- ActiveZero++: Mixed Domain Learning Stereo and Confidence-Based Depth Completion With Zero Annotation (TPAMI 2023 · similarity 0.97)
- ActiveStereoNet: End-to-End Self-Supervised Learning for Active Stereo Systems (ECCV 2018 · similarity 0.90)
- Connecting the Dots: Learning Representations for Active Monocular Depth Estimation (CVPR 2019 · similarity 0.89)
- UltraStereo: Efficient Learning-Based Matching for Active Stereo Systems (CVPR 2017 · similarity 0.88)
- Active Event-based Stereo Vision (CVPR 2025 · similarity 0.87)
- FoundationStereo: Zero-Shot Stereo Matching (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.
