Research map / Detection and segmentation / Open-set and few-shot detection
Robust Object Detection via Instance-Level Temporal Cycle Confusion
Xin Wang, Thomas E. Huang, Benlin Liu, Fisher Yu, Xiaolong Wang, Joseph E. Gonzalez, Trevor Darrell
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 CycConf auxiliary task | Faster R-CNN · BDD100K Daytime->Night | AP50: 31.35 -> 33.58 | Improved |
| add CycConf auxiliary task | Faster R-CNN · BDD100K Night->Daytime | AP50: 33.04 -> 34.34 | Improved |
| add CycConf auxiliary task | Faster R-CNN · Waymo Front Left->BDD100K Night | AP50: 19.62 -> 26.01 | Improved |
| add CycConf auxiliary task | Faster R-CNN · Waymo Front Right->BDD100K Night | AP50: 17.26 -> 20.58 | Improved |
| use cycle confusion (match most different instance) instead of cycle consistency (nearest neighbour) (ablation) | Faster R-CNN + CycConf · BDD100K Daytime->Night | AP50: 32.44 -> 33.58 | Improved |
| use cycle confusion instead of cycle consistency (ablation) | Faster R-CNN + CycConf · Waymo Front Left->BDD100K Night | AP50: 23.44 -> 26.01 | Improved |
| use cycle confusion instead of cycle consistency (ablation) | Faster R-CNN + CycConf · Waymo Front Right->BDD100K Night | AP50: 17.92 -> 20.58 | Improved |
| add image-level rotation task | Faster R-CNN · BDD100K Daytime->Night | AP50: 31.35 -> 32.95 | Improved |
| add Jigsaw task | Faster R-CNN · BDD100K Daytime->Night | AP50: 31.35 -> 31.22 | No clear effect |
| add Jigsaw task | Faster R-CNN · Waymo Front Right->BDD100K Night | AP50: 17.26 -> 16.58 | Hurt |
| add image rotation task with target-domain unlabeled data (UDA) | Faster R-CNN ResNet-101 · Cityscapes->Foggy Cityscapes | mAP50: 25.6 -> 41.5 | Improved |
| add image rotation task with target-domain unlabeled data (UDA) | Faster R-CNN VGG-16 · Cityscapes->Foggy Cityscapes | mAP50: 18.8 -> 37.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Faster R-CNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · better on 8 of 8 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
- Domain Adaptive Object Detection via Dual-Stream Bilevel-Cycle Optimization (ECCV 2026 · similarity 0.88)
- Instance-Aware, Context-Focused, and Memory-Efficient Weakly Supervised Object Detection (CVPR 2020 · similarity 0.88)
- Single-Domain Generalized Object Detection in Urban Scene via Cyclic-Disentangled Self-Distillation (CVPR 2022 · similarity 0.87)
- A Robust Learning Approach to Domain Adaptive Object Detection (ICCV 2019 · similarity 0.87)
- Object Discovery and Representation Networks (ECCV 2022 · similarity 0.86)
- Automatic Adaptation of Object Detectors to New Domains Using Self-Training (CVPR 2019 · 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.
