Research map / Detection and segmentation / 3D detection and driving
Test-Time Correction: An Online 3D Detection System via Visual Prompting
Hanxue Zhang, Zetong Yang, Yanan Sun, Li‐Qun Chen, Fei Xia, Fatma Güney, Hongyang Li
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 Test-time Correction with visual prompts at 5-frame feedback | BEVFormer · nuScenes val | mAP (%): 41.1 -> 50.2 | Improved |
| add Test-time Correction with visual prompts | StreamPETR · nuScenes val | mAP (%): 45.0 -> 51.6 | Improved |
| add Test-time Correction to the monocular detector | MonoDETR · nuScenes val | mAP (%): 42.6 -> 45.0 | Improved |
| add TTC to a detector trained with 20% distant labels | MonoDETR · nuScenes distant objects | mAP (%): 0.0 -> 10.1 | Improved |
| add TTC to a detector trained with 20% vehicle labels | MonoDETR · nuScenes vehicles | mAP (%): 2.4 -> 22.4 | Improved |
| add TTC for unseen classes (zero-shot) | MonoDETR · nuScenes unseen classes | mAP (%): 0.0 -> 7.2 | Improved |
| add TTC under rain and night domain shift | MonoDETR · nuScenes rain and night | mAP (%): 25.2 -> 29.0 | Improved |
| add explicit similarity loss (Focal and Dice) to alignment (ablation) | TTC-MonoDETR · nuScenes | mAP (%): 32.6 -> 39.4 | Improved |
| add localization loss on top of similarity loss (ablation) | TTC-MonoDETR · nuScenes | mAP (%): 39.4 -> 43.3 | Improved |
| predict N=4 positions per visual prompt instead of 1 (ablation) | TTC-MonoDETR · nuScenes | mAP (%): 39.9 -> 43.3 | Improved |
| increase N from 4 to 8 positions (ablation) | TTC-MonoDETR · nuScenes | mAP (%): 43.3 -> 43.0 | No clear effect |
| replace OA visual prompts with SOT tracking (ATOM) plus 3D head | MV2D · nuScenes val | mAP (%): 42.6 -> 42.1 | Hurt |
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.
- MonoDETR — MonoDETR: Depth-guided Transformer for Monocular 3D Object Detection (ICCV 2023) · better on 6 of 6 shared results
- BEVFormer — BEVFormer: Learning Bird’s-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers (ECCV 2022) · 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
- DriveGEN: Generalized and Robust 3D Detection in Driving via Controllable Text-to-Image Diffusion Generation (CVPR 2025 · similarity 0.85)
- OpenAD: Open-World Autonomous Driving Benchmark for 3D Object Detection (NeurIPS 2025 · similarity 0.85)
- Ev-3DOD: Pushing the Temporal Boundaries of 3D Object Detection with Event Cameras (CVPR 2025 · similarity 0.85)
- CodeMerge: Codebook-Guided Model Merging for Robust Test-Time Adaptation in Autonomous Driving (NeurIPS 2025 · similarity 0.85)
- RLGF: Reinforcement Learning with Geometric Feedback for Autonomous Driving Video Generation (NeurIPS 2025 · similarity 0.85)
- CorrBEV: Multi-View 3D Object Detection by Correlation Learning with Multi-modal Prototypes (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.
