@1/9/2026
주제: Prompt-controllable Camera Parameters in Image manipulation
1. Generative Photography: Scene-Consistent Camera Control for Realistic Text-to-Image Synthesis (CVPR 2025 Highlight)
2. Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation (ICLR submitted, Rate: 6,8,8,6)
발표자: 정성훈
주제 : Exploring 3D foundation models
1. DUSt3R: Geometric 3D Vision Made Easy [CVPR 2024]
2. VGGT: Visual Geometry Grounded Transformer [CVPR 2025 Best Paper Award]
발표자: 진인환
주제 : Deep Dive into Efficient LLMs: KV Cache Compression
1. SnapKV: LLM Knows What You are Looking for Before Generatio [NeurIPS 2024]
2. KVzip: Query-Agnostic KV Cache Compression with Context Reconstruction [NeurIPS 2025 oral]
발표자: 백창우
@1/15/2026
주제 : Agentic Query Refinement for Open-Vocabulary Object Detection
1. YOLOE: Real-Time Seeing Anything [ICCV 2025]
2. Re-Aligning Language to Visual Objects with an Agentic Workflow [ICLR 2025]
발표자: 정다운
주제: 3D Representation for Dynamic scenes
1. FreeTimeGS: Free Gaussian Primitives at Anytime and Anywhere for Dynamic Scene Reconstruction [CVPR2025]
2. 4DGT: Learning a 4D Gaussian Transformer Using Real-World Monocular Videos [NeurIPS2025 Spotlight]
발표자: 김태훈
주제: Introducing the Meta SAM Family
1. SAM 3: Segment Anything with Concepts [arXiv, 2025]
2. SAM Audio: Segment Anything in Audio [arXiv, 2025]
발표자: 박순찬
@1/23/2026
주제 : Controllable Video Generation with Physical Priors
1.
WonderPlay: Dynamic 3D Scene Generation from a Single Image and Actions [ICCV'25 Highlight]
2.
Force Prompting: Video Generation Models Can Learn and Generalize Physics-based Control Signals [NeurIPS'25]
발표자: 문형주
주제: real-to-sim-to-real pipeline for humanoid
1.
Visual Imitation Enables Contextual Humanoid Control (CoRL 2025, Best Student Paper Award.)
발표자: 이우진
주제 : Object hallucinations in VLM
1.
Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens (CVPR 2025)
2.
On Epistemic Uncertainty of Visual Tokens for Object Hallucinations in Large Vision-Language Models(NeurIPS 2025)
발표자: 김소현
@1/29/2026
주제: A Unified Feedforward Framework for Scene and Human Understanding
1. Continuous 3D Perception Model with Persistent State (CVPR 2025 Oral)
2. Human3R: Everyone Everywhere All at Once (ICLR 2026 submitted, Rate: 8, 8, 4)
발표자: 최지훈
주제: 딥러닝 홀로서기
발표자: 윤지원
주제: 딥러닝 홀로서기
발표자: 조범권