Source author record

Daiming Qu

Daiming Qu appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Artificial Intelligence Computer Vision Information Theory math.IT

Catalog footprint

What is connected

2works

4topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2021arXiv

LNSMM: Eye Gaze Estimation With Local Network Share Multiview Multitask

Eye gaze estimation has become increasingly significant in computer vision.In this paper,we systematically study the mainstream of eye gaze estimation methods,propose a novel methodology to estimate eye gaze points and eye gaze directions simultaneously.First,we construct a local sharing network for feature extraction of gaze points and gaze directions estimation,which can reduce network computational parameters and converge quickly;Second,we propose a Multiview Multitask Learning (MTL) framework,for gaze directions,a coplanar constraint is proposed for the left and right eyes,for gaze points,three views data input indirectly introduces eye position information,a cross-view pooling module is designed, propose joint loss which handle both gaze points and gaze directions estimation.Eventually,we collect a dataset to use of gaze points,which have three views to exist public dataset.The experiment show our method is state-of-the-art the current mainstream methods on two indicators of gaze points and gaze directions.

preprint2016arXiv

Improving Bandwidth Efficiency of FBMC-OQAM Through Virtual Symbols

Filter bank multicarrier (FBMC) systems that are based on offset quadrature amplitude modulation (OQAM), namely, FBMC-OQAM, have been criticized for their inefficiency in the use of spectral resources, because of the long ramp-up and ramp-down tails at the beginning and the end of each data packet, respectively. We propose a novel method for shortening these tails. By appending a set of virtual (i.e., none data carrying) symbols to the beginning and the end of each packet, and clever selection of these symbols, we show that the ramp-up and rampdown tails in FMBC-OQAM can be suppressed to an extent that they deem as negligible and thus may be ignored. This shortens the length of signal burst in each FBMC-OQAM packet, hence, improves on its bandwidth efficiency, viz., the same data is transmitted over a shorter period of time. We develop an optimization method that allows computation of virtual symbols, for each data packet. Simulation results show that, compared to existing methods, the proposed tail-shortening approach leads to a superior out-of-band (OOB) emission performance and a much lower error vector magnitude (EVM) for the demodulated symbols.