DAI-Net:一种新的低光照目标检测网络,通过零样本昼夜域适应来增强暗光目标检测,其目的是将检测器从明亮场景推广到暗光场景,而不需要真正的低光数据,性能表现SOTA!代码即将开源! 点击关注@CVer官方知乎账号,可以第一时间看到最优质、最前沿的CV、AI、AIGC工作~ DAI-Net Boosting Object Detection with Zero-Shot Da...
与现有的低光照图像增强和物体检测方法相比,DAI-Net具有以下优点: 无需低光照数据:DAI-Net不需要大量的低光照图像进行训练,这大大降低了数据收集和标注的成本。 泛化能力强:通过零样本学习,DAI-Net能够将日间场景的学习成果泛化到夜间场景,这在传统的域适应方法中是难以实现的。 性能提升显著:在多个数据集上的实验...
239 Noise-Aware Unsupervised Deep Lidar-Stereo Fusion https://github.com/AvrilCheng/LidarStereoNet Wednesday Poster 2.1 211 Xuelian Cheng Xuelian Cheng, Yiran Zhong, Yuchao Dai, Pan Ji, Hongdong Li 39 322 Handwriting Recognition in Low-Resource Scripts Using Adversarial Learning https://github.co...
3522 NM-Net: Mining Reliable Neighbors for Robust Feature Correspondences Chen Zhao (Huazhong University of Science and Technology); Zhiguo Cao (Huazhong Univ. of Sci.&Tech.); chi li (Huazhong University of Science and Technology); Xin Li (West Virginia University); Jiaqi Yang (Huazhong Univ....
总的来说,我们提出了一种新颖的方法,称为BiSRNet,用于在快照压缩成像(SCI)系统中从压缩测量中有效和实用地恢复HSI。我们重新设计了一个紧凑、易部署的基础模型,并提出了基本单元BiSR-Conv,用于自适应地重新分配HSI表示,并使用可缩放的双曲正切函数更准确地近似反向传播中的Sign函数。我们还定制了四个二值化卷积模块...
977 Scan2CAD: Learning CAD Model Alignment in RGB-D Scans Armen Avetisyan (Technical University of Munich)*; Manuel Dahnert (Technical University of Munich); Angela Dai (Technical University of Munich); Manolis Savva (Simon Fraser University); Angel X Chang (Eloquent Labs); Matthias Niessner (...
TransNeXt-Base ImageNet-1K 1x 51.7 45.9 109.2M model config log When we checked the training logs, we found that the mask mAP and other detailed performance of the Mask R-CNN using the TransNeXt-Tiny backbone were even better than reported in the paper (versions V1 and V2). We have ...
作者:Dai Shi(一位独立研究员) 代码:https://github.com/DaiShiResearch/TransNeXt 论文:https://arxiv.org/abs/2311.17132 CVPR 2024 论文和开源项目合集请戳—>https://github.com/amusi/CVPR2024-Papers-with-Code 由于残差连接的深度退化效应,许多依赖堆叠层进行信息交换的高效 Vision Transformers 模型往往无法...
Code: https://github.com/Dai-Wenxun/MotionLCMMotion Mamba: Efficient and Long Sequence Motion Generation with Hierarchical and Bidirectional Selective SSMPaper: https://arxiv.org/abs/2403.07487 Code: https://github.com/steve-zeyu-zhang/MotionMambaNeu...
HLNet, which secured fourth place in the NTIRE 2024 Challenge on Bracketing Image Restoration and Enhancement - Track 2 BracketIRE+ Task, has now been accepted by the CVPR 2024 Workshop. - chengeng0613/HLNet