EVA-CLIP combines pre-trained EVA, LAMB optimizer, and masking techniques to achieve 82.0% zero-shot ImageNet accuracy with 5.0B parameters, trained on 9B samples.

Quan Sun, Yuxin Fang, Ledell Wu et al.

WinCLIP achieves 91.8% AUROC in zero-shot anomaly classification on MVTec-AD.

Jongheon Jeong, Yang Zou, Taewan Kim et al.

Proposes OKDPH, a parameter hybridization framework, to flatten loss minima and enhance online knowledge distillation generalization.

Tianli Zhang, Mengqi Xue, Jiangtao Zhang et al.

YOSO introduces a real-time panoptic segmentation framework using dynamic convolution for unified instance and semantic masks, achieving 46.4 PQ and 45.6 FPS on COCO.

Jie Hu, Linyan Huang, Tianhe Ren et al.

DBARF self-supervises camera pose optimization using implicit cost functions, enhancing GeNeRF generalization.

Yu Chen, Gim Hee Lee

Make-It-3D leverages 2D diffusion priors with NeRF in a two-stage pipeline to produce high-fidelity 3D models from a single image, outperforming prior methods.

Junshu Tang, Tengfei Wang, Bo Zhang et al.

DreamBooth3D combines DreamBooth and DreamFusion to generate personalized 3D models from 3-6 images, achieving high-quality, editable assets.

Amit Raj, Srinivas Kaza, Ben Poole et al.

Zero-shot text-to-video synthesis leveraging Stable Diffusion with motion encoding and cross-frame attention, achieving high temporal consistency.

Levon Khachatryan, Andranik Movsisyan, Vahram Tadevosyan et al.

Proposes domain alignment and spatio-temporal regularization for knowledge transfer, boosting SNN generalization on event data.

Xiang He, Dongcheng Zhao, Yang Li et al.

NUWA-XL uses 'Diffusion over Diffusion' for ultra-long video generation, reducing inference time by 94.26%.

Shengming Yin, Chenfei Wu, Huan Yang et al.

Proposes a cascaded latent diffusion model with part-level implicit representation, achieving state-of-the-art 3D shape generation and editing without extra training.

Juil Koo, Seungwoo Yoo, Minh Hieu Nguyen et al.

Proposes CAT-Seg, a cost aggregation-based open-vocabulary semantic segmentation framework, fine-tuning CLIP encoders to improve pixel-level recognition of seen and unseen classes, outperforming state-of-the-art.

Seokju Cho, Heeseong Shin, Sunghwan Hong et al.

Text2Tex uses depth-aware diffusion models with auto view selection to synthesize high-quality 3D textures from text prompts.

Dave Zhenyu Chen, Yawar Siddiqui, Hsin-Ying Lee et al.

MM-REACT integrates ChatGPT with vision experts via prompts for advanced multimodal reasoning without training.

Zhengyuan Yang, Linjie Li, Jianfeng Wang et al.

Zero-1-to-3 leverages pre-trained diffusion models for zero-shot single-image 3D view synthesis and reconstruction, with strong generalization.

Ruoshi Liu, Rundi Wu, Basile Van Hoorick et al.

Proposes a decoupled encoder-decoder framework with StyleGAN for photorealistic image synthesis from abstract sketches, achieving state-of-the-art results.

Subhadeep Koley, Ayan Kumar Bhunia, Aneeshan Sain et al.

Introduced a foresightful dense visual affordance method, enhancing efficiency in deformable object manipulation.

Ruihai Wu, Chuanruo Ning, Hao Dong

HIVE leverages human feedback to enhance instruction-based visual editing via reward modeling and offline RL, significantly improving alignment.

Shu Zhang, Xinyi Yang, Yihao Feng et al.

Proposes Min-SNR-γ weighting strategy, accelerating diffusion model training by 3.4×, achieving FID 2.06 on ImageNet 256×256.

Tiankai Hang, Shuyang Gu, Chen Li et al.

Proposes DIRE, leveraging diffusion model reconstruction error for high-accuracy detection of diffusion-generated images, outperforming existing methods.

Zhendong Wang, Jianmin Bao, Wengang Zhou et al.