Created the first crowdsourced database of subjective optimal rate-distortion trade-offs for progressive image delivery. (Paper accepted at Mobile Multimedia Computing ICME 2017)
•
Applied machine learning (including deep learning) to detect visual events during website loading using computer vision techniques such as optical flow, motion history histograms, and histogram of oriented gradients etc.
•
Designed a responsive crowdsourcing web application to collect subjective rate-distortion trade-off opinions for baseline image delivery on different platforms in both full reference and no reference settings.
•
Moire pattern detection for WWW images using visual descriptors such as local binary pattern, DSIFT and IQA algorithms .
