GPU acceleration of NL-means, BM3D and VBM3D
Denoising is an essential part of any image- or video-processing pipeline. Unfortunately, due to time-processing constraints, many pipelines do not consider the use of modern denoisers. These algorithms have only CPU implementations or suboptimal GPU implementations. We propose a new efficient GPU implementation of NL-means and BM3D, and, to our knowledge, the first GPU implementation of the video-denoising algorithm VBM3D. The performance of these implementations enable their use in real-time scenarios.
Code (0)
등록된 구현이 없습니다.
Tasks
CPUDenoisingGPUVideo DenoisingSimilar Papers 제목 키워드 기반
Accelerating Spherical k-Means
Spherical k-means is a widely used clustering algorithm for sparse and high-dimensional data such as document vectors. While several improvements and accelerations have been introduced for the original k-means algorithm,…
ClusteringComputational EfficiencyFast K-Means Clustering with Anderson Acceleration
We propose a novel method to accelerate Lloyd's algorithm for K-Means clustering. Unlike previous acceleration approaches that reduce computational cost per iterations or improve initialization, our approach is focused o…
ClusteringExact Acceleration of K-Means++ and K-Means$\|$
K-Means++ and its distributed variant K-Means$\|$ have become de facto tools for selecting the initial seeds of K-means. While alternatives have been developed, the effectiveness, ease of implementation, and theoretical …
A Super Fast K-means for Indexing Vector Embeddings
We present SuperKMeans: a k-means variant designed for clustering collections of high-dimensional vector embeddings. SuperKMeans' clustering is up to 7x faster than FAISS and Scikit-Learn on modern CPUs and up to 4x fast…
QuicK-means: Acceleration of K-means by learning a fast transform
K-means -- and the celebrated Lloyd algorithm -- is more than the clustering method it was originally designed to be. It has indeed proven pivotal to help increase the speed of many machine learning and data analysis tec…
ClusteringData Compression