• Publications
  • Influence
Single Image Reflection Separation with Perceptual Losses
The approach uses a fully convolutional network trained end-to-end with losses that exploit low-level and high-level image information and proposes a novel exclusion loss that enforces pixel-level layer separation. Expand
Zoom to Learn, Learn to Zoom
This paper shows that when applying machine learning to digital zoom, it is beneficial to operate on real, RAW sensor data, and shows how to obtain such ground-truth data via optical zoom and contribute a dataset, SR-RAW, for real-world computational zoom. Expand
Portrait shadow manipulation
A computational approach is presented that gives casual photographers some of this control, thereby allowing poorly-lit portraits to be relit post-capture in a realistic and easily-controllable way and proposes a way to explicitly encode facial symmetry. Expand
Synthetic defocus and look-ahead autofocus for casual videography
A system that synthetically renders refocusable video from a deep DOF video shot with a smartphone, and analyzes future video frames to deliver context-aware autofocus for the current frame is presented. Expand
Photometric Stabilization for Fast‐forward Videos
A novel photometric stabilization algorithm for fast‐forward videos that is robust to large content‐variation across frames and that outperforms the state‐of‐the‐art on a new dataset consisting of controlled synthetic and real videos. Expand
Virtual fitting: real-time garment simulation for online shopping
An innovative approach is proposed that utilizes machine learning to meet this real-time requirement of garment simulation in real time and can be applied to virtual fitting systems. Expand
Time-Travel Rephotography
This paper simulates traveling back in time with a modern camera to rephotograph famous subjects, using the StyleGAN2 framework to project old photos into the space of modern high-resolution photos, achieving all of these effects in a unified framework. Expand
Single Image Reflection Separation with Perceptual Losses Supplementary Material
We illustrate how we generate our synthetic dataset. We followed a similar forward model described in CEILNet [1] with modification stated in Section 5.1 of the main paper. The vignette mask isExpand
Learned Dual-View Reflection Removal
The model consists of a learning-based reflection-invariant flow model for dual-view registration, and a learned synthesis model for combining aligned image pairs that outperforms existing single-image and multi-image dereflection approaches. Expand
Zoom to Learn, Learn to Zoom Supplementary Material
As detailed in Section 4, we analyze the percentage of features that are matched uniquely (i.e., bijectively) in nearest feature search when applying directly contextual loss to training. TheExpand