Z
Zezhou Cheng
Researcher at University of Massachusetts Amherst
Publications - 24
Citations - 771
Zezhou Cheng is an academic researcher from University of Massachusetts Amherst. The author has contributed to research in topics: Computer science & Invariant (mathematics). The author has an hindex of 5, co-authored 15 publications receiving 472 citations. Previous affiliations of Zezhou Cheng include Shanghai Jiao Tong University.
Papers
More filters
Proceedings ArticleDOI
Deep Colorization
TL;DR: Inspired by the recent success in deep learning techniques which provide amazing modeling of large-scale data, this paper re-formulates the colorization problem so thatDeep learning techniques can be directly employed and a joint bilateral filtering based post-processing step is proposed to ensure artifact-free quality.
Posted Content
Deep Colorization
TL;DR: Wang et al. as discussed by the authors re-formulated the colorization problem so that deep learning techniques can be directly employed and developed an adaptive image clustering technique to incorporate the global image information.
Proceedings ArticleDOI
A Bayesian Perspective on the Deep Image Prior
TL;DR: In this article, the authors show that the deep image prior is asymptotically equivalent to a stationary Gaussian process prior in the limit as the number of channels in each layer goes to infinity, and derive the corresponding kernel.
Proceedings ArticleDOI
A Realistic Evaluation of Semi-Supervised Learning for Fine-Grained Classification
TL;DR: In this article, the authors evaluate the effectiveness of semi-supervised learning on a realistic benchmark where data exhibits considerable class imbalance and contains images from novel classes, and find that standard fine-tuning followed by distillation-based self-training is the most robust.
Journal ArticleDOI
Colorization Using Neural Network Ensemble
TL;DR: This paper proposes a mixture learning model representing the presence of sub-color-style within an overall image data set and uses ensemble multiple neural networks to obtain better color estimation performance than could be obtained from any of the constituent neural network alone.