Research map / 3D vision / Depth and stereo
Towards Robust Driving Perception: A Flexible Scale-Driven Family for Self-Supervised Monocular Depth Estimation
Zhaowen Zhu, Li Zhang, Chen Yujie, Zhang Tian, Yingjie Wang, Mingxia Zhan
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 |
|---|---|---|---|
| use HPB post-convolution instead of HEB (ablation) | Flex-X-Large · KITTI Eigen split (640x192) | Sq Rel: 0.614 -> 0.605 | Improved |
| add dynamic upsampling instead of bilinear (ablation) | Flex-X-Large · KITTI Eigen split (640x192) | Sq Rel: 0.627 -> 0.605 | Improved |
| add inverted prediction head (ablation) | Flex-X-Large · KITTI Eigen split (640x192) | Sq Rel: 0.631 -> 0.605 | Improved |
| add stage-2 static-dynamic decoupled training (ablation) | Flex-X-Large · KITTI Eigen split (640x192) | Abs Rel: 0.100 -> 0.093 | Improved |
| use adaptive decoupled mask instead of fixed threshold (ablation) | Flex-X-Large · KITTI Eigen split (640x192) | Sq Rel: 0.629 -> 0.605 | Improved |
| add stage-2 dynamic masking instead of plain fine-tuning (ablation) | Flex-X-Large · Cityscapes | Abs Rel: 0.140 -> 0.086 | Improved |
| extend training with plain fine-tuning (Stage 1 + Fine-Tune) (ablation) | Flex-X-Large · KITTI | Abs Rel: 0.100 -> 0.102 | Hurt |
| replace MonoViT decoder with Ours X-Large decoder (ablation) | MonoViT · KITTI | FLOPs (G): 59.7 -> 33.2 | Improved |
| add stage-2 decoupled training to MonoViT + X-Large decoder (ablation) | MonoViT + Ours X-Large decoder · KITTI | Abs Rel: 0.099 -> 0.095 | Improved |
| add stage-2 decoupled training to Monodepth2 + Medium decoder (ablation) | Monodepth2 + Ours Medium/Large decoder · KITTI | Abs Rel: 0.112 -> 0.108 | Improved |
| add dynamic upsampling to 11l-seg encoder model (ablation) | Flex-Large · KITTI | RMSE: 4.272 -> 4.241 | Improved |
| use YOLO11n-seg encoder instead of ResNet-18 (ablation) | Monodepth2-style decoder · KITTI | Abs Rel: 0.115 -> 0.111 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GeoDepth — GeoDepth: From Point-to-Depth to Plane-to-Depth Modeling for Self-Supervised Monocular Depth Estimation (CVPR 2025) · better on 1 of 1 shared result
- ProDepth — ProDepth: Boosting Self-Supervised Multi-Frame Monocular Depth with Probabilistic Fusion (ECCV 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- The Road Ahead in Autonomous Driving: The KITScenes Multimodal Dataset (NeurIPS 2026 · similarity 0.87)
- Rethinking Lanes and Points in Complex Scenarios for Monocular 3D Lane Detection (CVPR 2025 · similarity 0.87)
- PPEA-Depth: Progressive Parameter-Efficient Adaptation for Self-Supervised Monocular Depth Estimation (AAAI 2024 · similarity 0.87)
- LaneSegNet: Map Learning with Lane Segment Perception for Autonomous Driving (ICLR 2024 · similarity 0.87)
- UrbanIng-V2X: A Large-Scale Multi-Vehicle, Multi-Infrastructure Dataset Across Multiple Intersections for Cooperative Perception (NeurIPS 2025 · similarity 0.87)
- RCooper: A Real-world Large-scale Dataset for Roadside Cooperative Perception (CVPR 2024 · similarity 0.87)
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.
