InterFuser uses Transformer-based sensor fusion to improve autonomous driving safety, achieving top CARLA leaderboard score of 76.18.

Hao Shao, Letian Wang, RuoBing Chen et al.

Utilizing CLIP for image perception assessment, achieving high correlation.

Jianyi Wang, Kelvin C. K. Chan, Chen Change Loy

Proposes MHST, a multi-modal transformer model integrating text, layout, and visual features, achieving significant improvements on the TAT-DQA dataset for complex document VQA.

Fengbin Zhu, Wenqiang Lei, Fuli Feng et al.

Proposes VAT with 4D Convolutional Swin Transformer for few-shot segmentation, achieving state-of-the-art results.

Sunghwan Hong, Seokju Cho, Jisu Nam et al.

RAVOS combines deep motion modeling with ROI prediction, achieving 86.1 J&F at 42 FPS for video object segmentation.

Bo Miao, Mohammed Bennamoun, Yongsheng Gao et al.

Proposes ST-P3, an end-to-end vision-based autonomous driving model using spatial-temporal features, achieving 85.2% mAP on nuScenes.

Shengchao Hu, Li Chen, Penghao Wu et al.

XMem employs a multi-store memory framework inspired by human cognition, achieving state-of-the-art long-video object segmentation with low GPU usage.

Ho Kei Cheng, Alexander G. Schwing

MSBReg combines delayed-parameter embeddings, multiview centroid, singular value regularization, and Brownian motion to improve self-supervised learning, achieving 81.56% top-1 accuracy on ImageNet.

Suhong Moon, Domas Buracas, Seunghyun Park et al.

Dual decision framework combining known class discriminator and class-agnostic object head improves open-set panoptic segmentation PQ by over 30%.

Hai-Ming Xu, Hao Chen, Lingqiao Liu et al.

Proposes an interpretable predictor framework using variational autoencoders and MCMC to select high-information queries, enhancing transparency.

Aditya Chattopadhyay, Stewart Slocum, Benjamin D. Haeffele et al.

Neural radiance fields (NeRF) combined with depth-guided ray marching enable high-fidelity 3D reconstruction of deformable tissues from single-view stereo videos, outperforming state-of-the-art methods.

Yuehao Wang, Yonghao Long, Siu Hin Fan et al.

Proposes HO-GCN with object dynamic descriptors for predicting human-large object interactions, achieving state-of-the-art accuracy and generalization.

Weilin Wan, Lei Yang, Lingjie Liu et al.

BEVDepth leverages explicit depth supervision to improve camera-based 3D detection, achieving 60.9% NDS on nuScenes.

Yinhao Li, Zheng Ge, Guanyi Yu et al.

VectorMapNet predicts sparse polylines end-to-end from onboard sensors, surpassing SOTA by 14.2 mAP on nuScenes.

Yicheng Liu, Tianyuan Yuan, Yue Wang et al.

UNIFIED-IO employs a unified sequence-to-sequence transformer architecture, converting multi-modal inputs and outputs into discrete tokens, trained on over 90 datasets, achieving state-of-the-art multi-task performance.

Jiasen Lu, Christopher Clark, Rowan Zellers et al.

Proposed TCP framework combines trajectory and control prediction, achieving 75.137 score on CARLA leaderboard with monocular input.

Penghao Wu, Xiaosong Jia, Li Chen et al.

RaMViD uses random-mask diffusion and reaches SOTA on BAIR/Kinetics-600.

Tobias Höppe, Arash Mehrjou, Stefan Bauer et al.

PIDNet introduces a three-branch architecture inspired by PID control, achieving real-time semantic segmentation with superior speed and accuracy.

Jiacong Xu, Zixiang Xiong, Shankar P. Bhattacharyya

This paper introduces a modular framework for diffusion models, employing high-order Runge-Kutta sampling to reduce NFE to 35 while achieving SOTA FID of 1.79 on CIFAR-10.

Tero Karras, Miika Aittala, Timo Aila et al.

TransFuser employs multi-scale Transformer fusion of camera and LiDAR data, achieving 20% higher driving scores and 48% fewer collisions in dense traffic scenarios.

Kashyap Chitta, Aditya Prakash, Bernhard Jaeger et al.