Huge congratulations to all our co-authors! 🎉 Six papers (5 main, 1 dataset) from MLV have been accepted to NeurIPS 2026! Check out our papers: mlv.korea.ac.kr/publications

We are presenting our work, “VideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinement,” at ECCV 2026 in Malmö! Time: Sep 10, 2026 • 4:30 PM – 6:30 PM CEST Location: Poster Session 2, ExHall, Poster #222 mlvlab.github.io/VideoSearch…

We are presenting our work, “Context Blindness in DPO: Mitigating Object Hallucination in MLLMs via Context-Calibrated Preference Optimization,” at ECCV 2026 in Malmö! Time: Sep 10, 2026 • 10:30 AM – 12:30 PM Poster Session 1, ExHall, Poster #467 arxiv.org/abs/2608.12158

We are presenting our work, “F⁴Splat: Feed-Forward Predictive Densification for Feed-Forward 3D Gaussian Splatting,” at ECCV 2026 in Malmö! Time: Sep 10, 2026 • 10:30 AM – 12:30 PM Location: Poster Session 1, ExHall, Poster #240 Project Page : mlvlab.github.io/F4Splat

We are presenting our work, “SVMemAgent: A Streaming Video Memory Agent for Query-Agnostic Online Frame Selection,” at the Wearables AI Workshop at ECCV 2026 in Malmö! Time: Sep 9, 2026 • 1:00 PM CEST Location: Malmö Arena Sportbaren We look forward to seeing you there!

We are presenting our work, “MoE-GRPO: Optimizing Mixture-of-Experts via Reinforcement Learning in Vision-Language Models” at CVPR 2026 in Denver! Time: Jun 6, 2026 • 11:45 AM – 1:45 PM Location: Poster Session 3, Hall F Paper: arxiv.org/abs/2603.24984

We are presenting our work, “RegFormer: Transferable Relational Grounding for Efficient Weakly-Supervised Human-Object Interaction Detection” at CVPR 2026 in Denver! Time: Jun 5, 2026 • 4:00 PM – 6:00 PM Location: Poster Session 2, Hall A-F Paper: arxiv.org/abs/2604.00507

We are presenting our work, “DocPrune: Efficient Document Question Answering via Background, Question, and Comprehension-aware Token Pruning” at CVPR 2026 in Denver! Time: Jun 5, 2026 • 10:45 AM – 12:45 PM Location: Poster Session 1, Hall A-F Paper: arxiv.org/abs/2604.22281

We are presenting our work, “SPRINT: Sparse-Dense Residual Fusion for Efficient Diffusion Transformers” at ICLR 2026 in Brazil! Time: Sat, Apr 25, 2026 • 3:15 PM – 5:45 PM Location: Pavilion 4 P4-3017 Paper: arxiv.org/abs/2510.21986 #ICLR2026 #Efficient #DiT #DiffusionModels

We are presenting our work, “Error as Signal: Stiffness-Aware Diffusion Sampling via Embedded Runge-Kutta Guidance” in Brazil, ICLR’26! Time: 4/23(Thu), 10:30 AM – 1:00 PM Location: Pavilion 4 P4-#3006 Paper: arxiv.org/abs/2603.03692 Video: youtu.be/E9i53cSxqDQ #ICLR2026

Huge congrats to all co-authors! Three papers from MLV got accepted to CVPR 2026! Please visit our homepage to take a closer look. mlv.korea.ac.kr/publications

🚀 Excited to share that we’ll be presenting our work, “DeepVideo-R1: Video Reinforcement Fine-Tuning via Difficulty-aware Regressive GRPO” at NeurIPS 2025 in San Diego! 🕙 When: 12/5 (Fri), 4:30 PM – 7:30 PM 📍 Where: Exhibit Hall C,D,E 🪧 Poster ID: 4418 See you soon!

🚀 Excited to share that we’ll be presenting our work, “Visual Diversity and Region-aware Prompt Learning for Zero-shot HOI Detection” at NeurIPS 2025 in San Diego! 🕙 When: 12/5 (Fri), 11:00 AM – 2:00 PM 📍 Where: Exhibit Hall C,D,E 🪧 Poster ID: 4911 See you soon!

😆 We presented our work "PRESTO: Preimage-Informed Instruction Optimization for Prompting Black-Box LLMs" at NeurIPS 2025 in San Diego! 🗓️ Date: 12/3 (Wed), 4:30 PM – 7:30 PM Paper: arxiv.org/pdf/2510.25808 We appreciate everyone who visited and discussed our work! #NeurIPS25

😆We presented our work "Blockwise Flow Matching: Improving Flow Matching Models For Efficient High-Quality Generation" at NeurIPS 2025 in San Diego 🗓️ Date: 12/3 (Wed), 11:00 AM – 2:00 PM Paper: arxiv.org/pdf/2510.21167 We appreciate everyone who visited and discussed our work!

😆 We are presenting our work, “Captioning for Text-Video Retrieval via Dual-Group Direct Preference Optimization” in Suzhou, EMNLP’25! @so0zip Paper: arxiv.org/pdf/2509.16560 Homepage: hyunwoojkim.com/publications See you soon!! #EMNLP2025 #Retrieval #VideoLLM #DPO

😆 We are presenting our work, “Representation Shift: Unifying Token Compression with FlashAttention” in Hawaii, ICCV’25! Time: 10/23(Thu), 14:15 Location: Exhibit Hall I PosterID: 1883 Paper: arxiv.org/pdf/2508.00367 #ICCV2025 #Efficiency #Representation

😆 We are presenting our work, “Bidirectional Likelihood Estimation with Multi-Modal Large Language Models for Text-Video Retrieval” in Hawaii, ICCV’25! Time:10/23(Thu), 10:45 Location: Exhibit Hall I PosterID: 232 Paper: arxiv.org/pdf/2507.23284 #ICCV2025 #Retrieval #MLLM

🎉 Four papers from MLV Lab have been accepted to NeurIPS 2025! 🎉 Huge congratulations to all the authors and collaborators on these amazing achievements @dogyunPark @parkjiny777 @Jwon_chu #NeurIPS2025 #MLVLab See you at San Diego & Mexico City! 🇺🇸🇲🇽

Congrats again! 🎉 One of our ICCV 2025 papers was selected as a Highlight! Big shoutout to the team “Bidirectional Likelihood Estimation with Multi-Modal LLMs for Text-Video Retrieval” @godohwan @so0zip @metaai #ICCV2025 #HighlightPaper #MLV #Multimodal #TextVideoRetrieval