Research map / Label-efficient and robust learning / Domain adaptation and generalization
Simulating Content Consistent Vehicle Datasets with Attribute Descent
Yue Yao, Liang Zheng, Xiaodong Yang, Milind Naphade, Tom Gedeon
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 SPGAN style domain adaptation to synthetic training data (ablation) | Attribute descent synthetic data · VehicleID | mAP: 24.36 -> 35.33 | Improved |
| add SPGAN style domain adaptation to synthetic training data (ablation) | Attribute descent synthetic data · VeRi-776 | mAP: 12.35 -> 21.29 | Improved |
| add second-stage finetuning on real data only (two-stage training) (ablation) | Joint training with attribute descent · VehicleID | mAP: 77.54 -> 81.35 | Improved |
| add second-stage finetuning on real data only (two-stage training) (ablation) | Joint training with attribute descent · VeRi-776 | mAP: 69.39 -> 70.62 | Improved |
| add second-stage finetuning on real data only (two-stage training) (ablation) | Joint training with attribute descent · CityFlow | mAP: 33.54 -> 37.16 | Improved |
| replace random attributes with attribute descent for synthetic-only training | VehicleX synthetic data · VehicleID (Large) | mAP: 22.00 -> 35.33 | Improved |
| replace random attributes with attribute descent for synthetic-only training | VehicleX synthetic data · VeRi-776 | Rank-1: +7.69% | Improved |
| replace random attributes with attribute descent in joint training | IDE joint training · VeRi-776 | mAP: 69.28 -> 70.62 | Improved |
| replace random attributes with attribute descent in joint training | IDE+Tri joint training · CityFlow | mAP: 35.96 -> 37.16 | Improved |
| add synthetic data with attribute descent to real training data | IDE · VehicleID (Large) | mAP: 78.51 -> 81.35 | Improved |
| replace random search with attribute descent for attribute optimization (ablation) | VehicleX attribute search · VehicleID (Large, synthetic only) | mAP: 26.35 -> 35.33 | Improved |
| replace LTS (FID-based) with attribute descent for attribute optimization (ablation) | VehicleX attribute search · VehicleID (Large, synthetic only) | mAP: 32.21 -> 35.33 | Improved |
Closest papers by meaning
- Attribute Descent: Simulating Object-Centric Datasets on the Content Level and Beyond (TPAMI 2023 · similarity 0.91)
- VehicleMAE: View-asymmetry Mutual Learning for Vehicle Re-identification Pre-training via Masked AutoEncoders (ICCV 2025 · similarity 0.85)
- Alice Benchmarks: Connecting Real World Re-Identification with the Synthetic (ICLR 2024 · similarity 0.85)
- Structural Information Guided Multimodal Pre-training for Vehicle-Centric Perception (AAAI 2024 · similarity 0.84)
- VQA-Diff: Exploiting VQA and Diffusion for Zero-Shot Image-to-3D Vehicle Asset Generation in Autonomous Driving (ECCV 2024 · similarity 0.83)
- DI-V2X: Learning Domain-Invariant Representation for Vehicle-Infrastructure Collaborative 3D Object Detection (AAAI 2024 · similarity 0.83)
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.
