Microsoft Research330 тыс
Опубликовано 18 августа 2016, 4:54
Since the early 1970s, computer vision researchers have relied on concepts from physics, mathematics, and statistics to develop new approaches for many computer vision problems. These include image formation models, regularization approaches, optimization techniques, Markov random fields, Bayesian inference, machine learning, manifold learning, and more recently, compressive sensing. In this talk, I will explore the notion that the latest excitement about compressive sensing and sparse representations is justified in the context of generating novel algorithms for computer vision problems. Examples from 3-D modeling from sparse gradients, dictionary-based face recognition, image reconstruction from gradients, and estimation of BRDFs will be provided to support the discussions.
Свежие видео
Случайные видео
Research intern talk: Real-time single-channel speech separation in noisy & reverberant environments