• 8:30 - 8:40 (10 min) Opening Remarks
• 8:40 - 9:20 (40 min) Keynote 1 - Dr. Liangqiong Qu (online or in-person, TBD) (35 min talk + 5 min Q&A)
• 9:20 - 9:45 (25 min) Oral Session 1 (2 papers; 10 min each incl. Q&A)
• 9:45 - 10:25 (40 min) Keynote 2 - Dr. Hao Wang (online) (35 min talk + 5 min Q&A)
• 10:25 - 10:40 (15 min) Coffee / Restroom Break
• 10:40 - 11:05 (25 min) Oral Session 2 (2 papers; 10 min each incl. Q&A)
• 11:05 - 11:45 (40 min) Keynote 3 - Dr. Ang Li (online) (35 min talk + 5 min Q&A)
• 11:45 - 12:00 (15 min) Best Paper Award + Closing

The following papers were selected for oral presentation at the 5th Federated Learning for Computer Vision (FedVision) Workshop at CVPR 2026.

Oral Session 1 — 9:20 AM–9:45 AM

1. Data-Free Contribution Estimation in Federated Learning using Gradient von Neumann Entropy
   (Best Paper Award Candidate)
2. FedSIR: Spectral Client Identification and Relabeling for Federated Learning with Noisy Labels
   (Best Paper Award Candidate)

Oral Session 2 — 10:40 AM–11:05 AM

3. UniFed-LoRA: Exploiting Semantic Task Correlation for Heterogeneous Multimodal Federated Fine-Tuning
   (Best Paper Award Candidate)
4. Conditional Generation for Missing Modality Imputation in Multi-Modal Federated Learning
   (Best Paper Award Candidate)

Each oral presentation is allocated 10 minutes, including Q&A.

In addition to the oral presentations, all accepted papers will be featured as poster presentations during the workshop.

Poster Session Time: 1:00 PM - 3:00 PM

Location: Exhibit Hall A

Poster Board Numbers Assigned to FedVision: 346-352

Please note that each poster board face is shared by two posters. We encourage authors to arrive early to locate their assigned board and set up their posters before the session begins.