Research map / Detection and segmentation / Open-set and few-shot detection
Debiased Teacher for Day-to-Night Domain Adaptive Object Detection
Yiming Cui, Liang Li, Haibing Yin, Yuhan Gao, Yaoqi Sun, Chenggang Yan
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 noise night modeling (NNM) to illumination modeling in DNDT (ablation) | DeT DNDT · BDD100k | mAP: 46.6 -> 48.2 | Improved |
| add light flare modeling (LFM) to DNDT (ablation) | DeT DNDT · TDND | mAP: 37.1 -> 38.7 | Improved |
| add full DNDT module to baseline (ablation) | Faster R-CNN self-training baseline · BDD100k | mAP: 45.8 -> 48.9 | Improved |
| add bidirectional mixing without ICFB (ablation) | DNDT model · TDND | mAP: 38.7 -> 37.6 | Hurt |
| add ICFB to bidirectional mixing (full CDRC) (ablation) | DNDT + Bi-Mix · TDND | mAP: 37.6 -> 44.8 | Improved |
| add ConCal (LPS and HPD) on top of CDRC (ablation) | DNDT + CDRC · TDND | mAP: 44.8 -> 46.7 | Improved |
| use night-like images rather than target images for class confidence statistics (ablation) | ConCal · SHIFT | mAP: 50.9 -> 52.6 | Improved |
| mix with night-like images instead of source images in CDRC (ablation) | CDRC · TDND | mAP: 41.7 -> 46.7 | Improved |
| replace CDRC with Mosaic mixing (ablation) | DeT · TDND | mAP: -5.7 | Hurt |
| add unidirectional D2N mixing with ICFB (ablation) | DeT · TDND | mAP: 43.4 -> 46.3 | Improved |
| replace CDRC with Mix-up (ablation) | DeT · BDD100k | mAP: -0.3 | Hurt |
| vary balance factor alpha from 1 to 5 (ablation) | CDRC · TDND | mAP: 45.3 -> 46.7 -> 44.8 | Mixed |
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.
- ISP-Teacher — ISP-Teacher:Image Signal Process with Disentanglement Regularization for Unsupervised Domain Adaptive Dark Object Detection (AAAI 2024) · better on 3 of 3 shared results
- 2PCNet — 2PCNet: Two-Phase Consistency Training for Day-to-Night Unsupervised Domain Adaptive Object Detection (CVPR 2023) · better on 3 of 3 shared results
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
- Boosting Object Detection with Zero-Shot Day-Night Domain Adaptation (CVPR 2024 · similarity 0.86)
- NightCC: Nighttime Color Constancy via Adaptive Channel Masking (CVPR 2024 · similarity 0.86)
- Day-Night Cross-domain Vehicle Re-identification (CVPR 2024 · similarity 0.86)
- DNIT: Enhancing Day-Night Image-to-Image Translation through Fine-Grained Feature Handling (Student Abstract) (AAAI 2024 · similarity 0.85)
- NightRain: Nighttime Video Deraining via Adaptive-Rain-Removal and Adaptive-Correction (AAAI 2024 · similarity 0.85)
- ISP-Teacher:Image Signal Process with Disentanglement Regularization for Unsupervised Domain Adaptive Dark Object Detection (AAAI 2024 · 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.
