整理：AI算法与图像处理
CVPR2022论文和代码整理：https://github.com/DWCTOD/CVPR2022-Papers-with-Code-Demo
ECCV2022论文和代码整理：https://github.com/DWCTOD/ECCV2022-Papers-with-Code-Demo
最新成果demo展示：
标题：
HDR-Plenoxels: Self-Calibrating High Dynamic Range Radiance Fields
论文：https://arxiv.org/abs/2208.06787
代码：https://github.com/postech-ami/HDR-Plenoxels
摘要：
提出了高动态范围辐射（HDR）场（HDR Plenoxels），它可以学习3D HDR辐射场的全光函数、几何信息以及2D低动态范围（LDR）图像中固有的不同相机设置。我们基于体素的体绘制管道仅使用端到端的方式从不同相机设置中获取的多视图LDR图像重建HDR辐射场，并且具有快速收敛速度。为了处理现实场景中的各种相机，我们引入了一个色调映射模块，该模块对数码相机成像管道（ISP）进行建模，并解开辐射设置的纠缠。我们的色调映射模块允许我们通过控制每个新奇视图的辐射设置进行渲染。最后，我们构建了一个具有不同相机条件的多视图数据集，它符合我们的问题设置。我们的实验表明，HDR Plenoxels可以从带有各种相机的LDR图像中表达细节和高质量的HDR新颖视图。
最新论文整理
ECCV2022
Updated on : 4 Oct 2022
total number : 2
From Face to Natural Image: Learning Real Degradation for Blind Image Super-Resolution
- 论文/Paper: http://arxiv.org/pdf/2210.00752
- 代码/Code: https://github.com/csxmli2016/redegnet
Long-Tailed Class Incremental Learning
- 论文/Paper: http://arxiv.org/pdf/2210.00266
- 代码/Code: https://github.com/xialeiliu/long-tailed-cil
CVPR2022
NeurIPS
Updated on : 4 Oct 2022
total number : 10
Expediting Large-Scale Vision Transformer for Dense Prediction without Fine-tuning
- 论文/Paper: http://arxiv.org/pdf/2210.01035
- 代码/Code: None
Perceptual Attacks of No-Reference Image Quality Models with Human-in-the-Loop
- 论文/Paper: http://arxiv.org/pdf/2210.00933
- 代码/Code: None
Learning Equivariant Segmentation with Instance-Unique Querying
- 论文/Paper: http://arxiv.org/pdf/2210.00911
- 代码/Code: https://github.com/jamesliang819/instance_unique_querying
Heatmap Distribution Matching for Human Pose Estimation
- 论文/Paper: http://arxiv.org/pdf/2210.00740
- 代码/Code: None
Unsupervised Multi-View Object Segmentation Using Radiance Field Propagation
- 论文/Paper: http://arxiv.org/pdf/2210.00489
- 代码/Code: None
Alignment-guided Temporal Attention for Video Action Recognition
- 论文/Paper: http://arxiv.org/pdf/2210.00132
- 代码/Code: None
An In-depth Study of Stochastic Backpropagation
- 论文/Paper: http://arxiv.org/pdf/2210.00129
- 代码/Code: None
Untargeted Backdoor Watermark: Towards Harmless and Stealthy Dataset Copyright Protection
- 论文/Paper: http://arxiv.org/pdf/2210.00875
- 代码/Code: https://github.com/thuyimingli/untargeted_backdoor_watermark
Compositional Generalization in Unsupervised Compositional Representation Learning: A Study on Disentanglement and Emergent Language
- 论文/Paper: http://arxiv.org/pdf/2210.00482
- 代码/Code: None
MaskTune: Mitigating Spurious Correlations by Forcing to Explore
- 论文/Paper: http://arxiv.org/pdf/2210.00055
- 代码/Code: https://github.com/aliasgharkhani/masktune
