Research map / Label-efficient and robust learning / Domain adaptation and generalization
Unsupervised Monocular Depth Estimation for Night-time Images using Adversarial Domain Feature Adaptation
Madhu Vankadari, Sourav Garg, Anima Majumder, Swagat Kumar, Ardhendu Behera
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 |
|---|---|---|---|
| apply ADFA to a day-trained Monodepth2 instead of training Monodepth2 on night images | Monodepth2 · Oxford RobotCar night-time (60m) | Abs Rel: 0.3990 -> 0.2327 | Improved |
| apply ADFA instead of Cycle-GAN image translation | day-time depth network · Oxford RobotCar night-time (60m) | Abs Rel: 0.7587 -> 0.2327 | Improved |
| apply ADFA instead of a day-trained Monodepth2 tested at night | Monodepth2 · Oxford RobotCar night-time (60m) | RMSE: 14.253 -> 10.089 | Improved |
| pretrain the day model on Oxford instead of KITTI (ablation) | ADFA · Oxford RobotCar night-time (60m) | Abs Rel: 0.3589 -> 0.2327 | Improved |
| pretrain the day model on Oxford instead of KITTI (ablation) | ADFA · Oxford RobotCar night-time (40m) | delta<1.25: 0.495 -> 0.735 | Improved |
| skip the first conv layer's day-encoder features in adversarial learning (ablation) | ADFA · Oxford RobotCar night-time (40m) | Abs Rel: 0.2071 -> 0.2038 | Improved |
| skip the first two conv layers' day-encoder features (ablation) | ADFA · Oxford RobotCar night-time (40m) | Abs Rel: 0.2071 -> 0.2005 | Improved |
| skip the first three conv layers' day-encoder features (ablation) | ADFA · Oxford RobotCar night-time (40m) | Abs Rel: 0.2005 -> 0.2260 | Hurt |
| skip the first three instead of two conv layers' features (ablation) | ADFA · Oxford RobotCar night-time (40m) | Sq Rel: 2.575 -> 2.574 | No clear effect |
| use Cycle-GAN translation instead of training Monodepth2 on night images | Monodepth2 · Oxford RobotCar night-time (60m) | Abs Rel: 0.3990 -> 0.7587 | Hurt |
Closest papers by meaning
- Self-Supervised Monocular Depth Estimation for All Day Images Using Domain Separation (ICCV 2021 · similarity 0.88)
- DeFeat-Net: General Monocular Depth via Simultaneous Unsupervised Representation Learning (CVPR 2020 · similarity 0.88)
- A One-Stage Domain Adaptation Network With Image Alignment for Unsupervised Nighttime Semantic Segmentation (TPAMI 2021 · similarity 0.87)
- Exploring the Common Appearance-Boundary Adaptation for Nighttime Optical Flow (ICLR 2024 · similarity 0.85)
- NightCC: Nighttime Color Constancy via Adaptive Channel Masking (CVPR 2024 · similarity 0.84)
- NightAdapter: Learning a Frequency Adapter for Generalizable Night-time Scene Segmentation (CVPR 2025 · similarity 0.84)
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.
