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
标题:SCAM! Transferring humans between images with Semantic Cross Attention Modulation
主页:https://imagine.enpc.fr/~dufourn/publications/scam.html
代码:https://github.com/nicolas-dufour/SCAM
论文:https://arxiv.org/pdf/2210.04883v1.pdf
最近的大量工作以语义条件下的图像生成为目标。大多数这类方法只关注较窄的姿势转移任务,而忽略了更具挑战性的对象转移任务,即不仅转移姿势,还转移外观和背景。在这项工作中,我们引入了SCAM(Semantic Cross Attention Modulation,语义交叉注意调制),这是一个系统,它对图像的每个语义区域(包括前景和背景)中丰富多样的信息进行编码,从而实现了以细节为重点的精确生成。这是由Semantic Attention Transformer Encoder实现的,该编码器为每个语义区域提取多个潜在向量,以及通过使用语义交叉注意调制来利用这些潜在向量的相应生成器。它仅使用重建设置进行训练,而受试者在测试时进行转移。我们的分析表明,我们提出的架构在编码每个语义区域的外观多样性方面是成功的。iDesigner和CelebAMask HD数据集上的大量实验表明,SCAM优于SEAN和SPADE;此外,它还开创了学科转移的新境界。
Improving the Reliability for Confidence Estimation
OpenOOD: Benchmarking Generalized Out-of-Distribution Detection
HSurf-Net: Normal Estimation for 3D Point Clouds by Learning Hyper Surfaces
RTFormer: Efficient Design for Real-Time Semantic Segmentation with Transformer
LION: Latent Point Diffusion Models for 3D Shape Generation
SageMix: Saliency-Guided Mixup for Point Clouds
Feature-Proxy Transformer for Few-Shot Segmentation
Adv-Attribute: Inconspicuous and Transferable Adversarial Attack on Face Recognition
Scalable Neural Video Representations with Learnable Positional Features
ALIFE: Adaptive Logit Regularizer and Feature Replay for Incremental Semantic Segmentation
Intermediate Prototype Mining Transformer for Few-Shot Semantic Segmentation
Q-ViT: Accurate and Fully Quantized Low-bit Vision Transformer
Structural Pruning via Latency-Saliency Knapsack
S4ND: Modeling Images and Videos as Multidimensional Signals Using State Spaces
Task-Free Continual Learning via Online Discrepancy Distance Learning
Flare7K: A Phenomenological Nighttime Flare Removal Dataset
PDEBENCH: An Extensive Benchmark for Scientific Machine Learning
Brain Network Transformer