Recovery of the Trajectory of DNA inside Viral Capsids from Liquid Crystal Vector Fields
Bacteriophages are of great interest for their potential in biotechnology and medical use, but the folding of DNA inside the capsids is unknown. My work connects the energy fields proposed for modeling DNA and the topological properties of phages.
A complete trajectory of DNA inside bacteriophage capsids using all information from the energy fields. But this one is too complicate to understand the structure ( more than 15k vertices and 300k crossings if projected to a plane ).
Results of sampling 5,000 (left) and 50,000 (right) vectors from the energy fields (with total approximately 200,000).
Result of generating microvectors (blue) around the macrovectors (black) with different restrictions using nonlinear optimization. Here all microvectors have norm = 1 and the mean of them is the macrovector.
Undergraduate Thesis
thesis.pdf
Talk slides
thesis_presentation_Zhijie.pdf
Deep Learning based Approach in Detecting Electrical Safety Harness
Traditional supervision method is visual monitoring by engineers which failed occasionally because managers cannot check every hour and day. This deep learning approach allow engineers to monitor through CCTV and in-person with the help of smart helmet products by Foshan Victor Electrical Equipment Co., Ltd.
Some successful detection of wearing electrical safety harness (blue only). Green boxes indicate harness detection while blue boxes not.