StreamMem employs query-agnostic KV cache compression, enabling efficient long video understanding with fixed memory, outperforming state-of-the-art methods.

Yanlai Yang, Zhuokai Zhao, Satya Narayan Shukla et al.

GeoSAM2 achieves 84.06% mIoU on PartObjaverse-Tiny and 74.42% on PartNetE using multi-view 2D mask prediction for 3D segmentation.

Ken Deng, Yunhan Yang, Jingxiang Sun et al.

PhysGM predicts 3D Gaussian and physical parameters from a single image, enabling real-time 4D scene synthesis in under one minute.

Chunji Lv, Zequn Chen, Donglin Di et al.

EgoTwin uses a diffusion transformer to generate consistent first-person video and human motion.

Jingqiao Xiu, Fangzhou Hong, Yicong Li et al.

DyCrowd framework achieves spatio-temporal consistent 3D reconstruction of dynamic crowds from large-scene videos.

Hao Wen, Hongbo Kang, Jian Ma et al.

TiP4GEN employs a dual-branch diffusion framework with 3D Gaussian Splatting for high-quality dynamic panoramic 4D scene synthesis.

Ke Xing, Hanwen Liang, Dejia Xu et al.

EgoLoc employs hand 3D dynamics and vision-language models for zero-shot temporal interaction localization, precisely identifying contact and separation moments.

Junyi Ma, Erhang Zhang, Yin-Dong Zheng et al.

Thyme integrates code-driven image operations and reasoning, boosting multimodal model performance.

Yi-Fan Zhang, Xingyu Lu, Shukang Yin et al.

Proposes TTF-VLA, a training-free temporal fusion method combining pixel difference and attention, boosting robotic manipulation success by 4-8%.

Chenghao Liu, Jiachen Zhang, Chengxuan Li et al.

TexVerse provides 1.6M high-resolution 3D models, addressing gaps in texture generation datasets.

Yibo Zhang, Li Zhang, Rui Ma et al.

Enhancing Multimodal LLMs with external tools across four key dimensions.

Wenbin An, Jiahao Nie, Yaqiang Wu et al.

Echo-4o leverages GPT-4o synthetic images to enhance image generation, achieving significant performance improvements across benchmarks.

Junyan Ye, Dongzhi Jiang, Zihao Wang et al.

Proposed M3-Agent with entity-centric multimodal long-term memory, outperforming baselines by 6-8% on long-video QA benchmarks.

Lin Long, Yichen He, Wentao Ye et al.

CitySeg uses local-global cross-attention and hierarchical classification to achieve open-vocabulary semantic segmentation and zero-shot inference for city-scale point clouds.

Jialei Xu, Zizhuang Wei, Weikang You et al.

Proposes DPGNet, leveraging unlabeled fake face data with cross-modal alignment and curriculum pseudo-labeling, boosting detection accuracy by 4.2%.

Zhiqiang Yang, Renshuai Tao, Chunjie Zhang et al.

ReconDreamer-RL integrates diffusion priors for scene reconstruction, reducing collision rate from 0.386 to 0.077 in autonomous driving training.

Chaojun Ni, Guosheng Zhao, Xiaofeng Wang et al.

LaRender renders object latents with volumetric compositing, reaching 0.767 human-perceived occlusion success on RealOcc.

Xiaohang Zhan, Dingming Liu

Proposes SOFA, a deep learning framework that simulates post-ablation scars, predicts AF recurrence risk, and optimizes procedural parameters, reducing predicted risk by 22.18%.

Yunsung Chung, Chanho Lim, Ghassan Bidaoui et al.

Restage4D leverages single video and motion priors to achieve physically consistent 4D deformable scene reconstruction with geometric correction.

Jixuan He, Chieh Hubert Lin, Lu Qi et al.

GenD fine-tunes only LayerNorm parameters, achieving state-of-the-art cross-dataset AUROC in deepfake detection, surpassing complex models.

Andrii Yermakov, Jan Cech, Jiri Matas et al.