Research map / Vision-language and multimodal / Cross-modal learning
FusionAgent: A Multimodal Agent with Dynamic Model Selection for Human Recognition
Jie Zhu, Xiao Guo, Yiyang Su, Anil Jain, Xiaoming Liu
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 Z-score confidence weighting to ACT (ablation) | FusionAgent ACT · LTCC | FNIR@1%FPIR: 62.4 -> 51.0 | Improved |
| use Z-score instead of min-max confidence weights (ablation) | FusionAgent ACT · LTCC | FNIR@1%FPIR: 60.8 -> 51.0 | Improved |
| add top-k selection to ACT (ablation) | FusionAgent ACT · CCVID | TAR@1%FAR: 83.3 -> 85.8 | Improved |
| add top-k selection to ACT (ablation) | FusionAgent ACT · MEVID | Rank-1: 54.1 -> 54.7 | No clear effect |
| add top-k selection to ACT (ablation) | FusionAgent ACT · LTCC | Rank-1: 75.1 -> 75.5 | No clear effect |
| use agent-selected models instead of hard selection with all models (ablation) | FusionAgent · LTCC | Rank-1, mAP, TAR | Improved |
| use ACT instead of Z-score fusion on agent-selected models (ablation) | FusionAgent · LTCC | FNIR@1%FPIR: 63.7 -> 51.0 | Improved |
| use ACT instead of FarSight fusion on agent-selected models (ablation) | FusionAgent · LTCC | FNIR@1%FPIR: 62.5 -> 51.0 | Improved |
| use agent selection with Z-score fusion instead of QME (ablation) | FusionAgent · LTCC | Rank-1: 73.8 -> 74.8 | Improved |
| use agent-selected combination instead of dataset-level grid-searched AIM+CAL (ablation) | FusionAgent · LTCC | FNIR@1%FPIR: 53.4 -> 51.0 | Improved |
| use direct answering instead of chain-of-thought (ablation) | FusionAgent · CCVID | Rank-1: 93.4 -> 92.8 | Mixed |
| fine-tune on 10-shot LTCC after CCVID training (ablation) | FusionAgent · CCVID to LTCC | Rank-1: 60.4 -> 73.6 | Improved |
Closest papers by meaning
- SapiensID: Foundation for Human Recognition (CVPR 2025 · similarity 0.85)
- FuseMoE: Mixture-of-Experts Transformers for Fleximodal Fusion (NeurIPS 2024 · similarity 0.84)
- Flexible Biometrics Recognition: Bridging the Multimodality Gap through Attention Alignment and Prompt Tuning (CVPR 2024 · similarity 0.84)
- Mordal: Automated Pretrained Model Selection for Vision Language Models (ICLR 2026 · similarity 0.83)
- SapiensID 2.0: Aligning Human Recognition Foundation Models with Human Perception (NeurIPS 2026 · similarity 0.83)
- MAT-Agent: Adaptive Multi-Agent Training Optimization (NeurIPS 2025 · similarity 0.82)
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.
