Dr. Qifeng CHEN Received Google Faculty Research Awards 2018

Dr. Qifeng CHEN, Assistant Professor of CSE and ECE, has been the Google Faculty Research Awards 2018 in Machine Perception.

Google Faculty Research Awards provides funds as a support for potential research at institutions and universities around the world. The program focuses on funding world-class technical research in Computer Science, Engineering, and related fields. For this year's award, there were 910 proposals from 40 countries and 320 countries, and including Dr. Chen's proposal, only 158 proposals were selected.

Dr. Chen's project, "Learning to see in the dark" aims to solve imaging problems taken in the dark. Due to the low photon count and low SNR, imaging in low light is challenging. Short-exposure images suffer from noise, while long exposure can induce blue and is often impractical. A variety of denoising, deblurring, and enhancement techniques have been proposed, but their effectiveness is limited in extreme conditions, such as video-rate imaging at night. To support the development of learning-based pipelines for low-light image processing, Dr. Chen and his team introduce a dataset of raw short-exposure low-light images, with corresponding long-exposure reference images. Using the presented dataset, they developed a pipeline for processing low-light images, based on end-to-end training of a fully-convolutional network. The network operates directly on raw sensor data and replaces much of the traditional image processing pipeline, which tends to perform poorly on such data. They reported promising results on the new dataset, analyze factors that affect performance, and highlighted opportunities for future work.

(a) An image captured at night by the Fujifilm X-T2 camera with ISO 800, aperture f/7.1, and exposure of 1/30 second. The illuminance at the camera is approximately 1 lux. (b) Processing the raw data by a traditional pipeline does not effectively handle the noise and color bias in the data. (c) Dr. Chen's result obtained from the same raw data.

(a) An image captured at night by the Fujifilm X-T2 camera with ISO 800, aperture f/7.1, and exposure of 1/30 second. The illuminance at the camera is approximately 1 lux. (b) Processing the raw data by a traditional pipeline does not effectively handle the noise and color bias in the data. (c) Dr. Chen's result obtained from the same raw data.

Thanks to Dr. Chen's research, problems encountered from imaging in low light could now be solved. Congratulations to Dr. Chen!

For more information, please visit the official announcement.