Research map / 3D vision / Depth and stereo
SaccadeCam: Adaptive Visual Attention for Monocular Depth Sensing
Brevin Tilmon, Sanjeev J. Koppal
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 fovea weighting of the photometric loss to SaccadeCam training (ablation) | SaccadeCam · KITTI depth, bandwidth (c) target 8 px/mm | RMSE: 6.815 -> 6.555 | Improved |
| add fovea weighting of the photometric loss to SaccadeCam training (ablation) | SaccadeCam · KITTI depth, bandwidth (a) target 35 px/mm | Abs Rel: 0.115 -> 0.116 | No clear effect |
| add fovea weighting of the photometric loss to SaccadeCam training (ablation) | SaccadeCam · KITTI depth, bandwidth (b) target 27 px/mm | Abs Rel: 0.121 -> 0.119 | Improved |
| learn adaptive resolution distribution (SaccadeCam) instead of uniform target resolution | self-supervised depth network · KITTI depth, bandwidth (c) target 8 px/mm | Abs Rel: 0.194 -> 0.164 | Improved |
| learn adaptive resolution distribution (SaccadeCam) instead of uniform target resolution | self-supervised depth network · KITTI depth, bandwidth (a) target 35 px/mm | Abs Rel: 0.117 -> 0.115 | Improved |
| learn adaptive resolution distribution (SaccadeCam) instead of uniform target resolution | self-supervised depth network · KITTI depth, bandwidth (b) target 27 px/mm | Abs Rel: 0.118 -> 0.119 | No clear effect |
| use learned attention instead of color edge detector as attention proxy (ablation) | SaccadeCam · KITTI depth, bandwidth (b) target 27 px/mm | Abs Rel: 0.137 -> 0.121 | Improved |
| use learned attention instead of color edge detector as attention proxy (ablation) | SaccadeCam · KITTI depth, bandwidth (c) target 8 px/mm | Abs Rel: 0.167 -> 0.167 | No clear effect |
| place focused depth in oracle-selected high-error regions of WAC image (True Oracle) | Wide angle camera depth · KITTI oracle study (c) | Abs Rel: 0.128 -> 0.112 | Improved |
| choose attention regions by photometric error instead of error against LIDAR (True Oracle) (ablation) | Oracle SaccadeCam · KITTI oracle study (a) | Abs Rel: 0.114 -> 0.116 | Hurt |
| increase full resolution to 70 px/mm from target 31 px/mm | monocular depth network · KITTI depth, full resolution 70 px/mm | Abs Rel: 0.118 -> 0.109 | Improved |
Closest papers by meaning
- Improved Efficiency Based on Learned Saccade and Continuous Scene Reconstruction From Foveated Visual Sampling (ICLR 2024 · similarity 0.88)
- Self-Supervised Monocular Trained Depth Estimation Using Self-Attention and Discrete Disparity Volume (CVPR 2020 · similarity 0.86)
- The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth (CVPR 2021 · similarity 0.86)
- Digging Into Self-Supervised Monocular Depth Estimation (ICCV 2019 · similarity 0.86)
- Generating and Exploiting Probabilistic Monocular Depth Estimates (CVPR 2020 · similarity 0.86)
- DepthCues: Evaluating Monocular Depth Perception in Large Vision Models (CVPR 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.
