http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Concurrent Food Localization and Recognition using Deep Convolution Neural Network
Rohit Thakur(로힛 타쿠르),Jusung Kang(강주성),Hyunjun Han(한현준),Heung-No Lee(이흥노) 대한전자공학회 2017 대한전자공학회 학술대회 Vol.2017 No.6
The analysis of food quality is considered as an important task because of the people’s nutrition habits that affect the global population. For this, it is very important to keep track of our eating habits by making automatic nutrition diaries. Keeping this in mind, we demonstrate how a deep learning technique is used to the tasks of localizing and recognizing various types of foods present in given images. Firstly, the Grad-CAM algorithm (a class-discriminative localization technique) along with data processing is used for generating discriminative image regions which helps in obtaining object proposals for detection purpose. Secondly, a classifier for object recognition is employed over it. Further, we compare our result with best state of art work and show our proposed method to be well equipped in terms of precision and accuracy. Through experiments we have verified that proposed approach is a convincing solution for concurrent food localization and recognition.
Rohit Kumar,Anita Sudhaik,Aftab Aslam Parwaz Khan,Pankaj Raizada,Abdullah M. Asiri,Satyabrata Mohapatra,Sourbh Thakur,Vijay Kumar Thakur,Pardeep Singh 한국공업화학회 2022 Journal of Industrial and Engineering Chemistry Vol.106 No.-
Growing pollution and high energy demand are some of the major issues against humans. Waterpollution is one of the main problems created due to industrial waste and irresponsible human activities. Nowadays, photocatalysis is rising as an efficient alternative to overcome the energy requirement andwater purification, as it can use solar light as a source of energy. Due to some limitations of photocatalysts,such as photocarriers recombination, low surface area, and limited light absorption, modificationssuch as defect modifications, heterojunction can be employed. Out of all, Z-scheme heterojunction is aproficient way to overwhelm the limitation of pristine photocatalysts. To enhance the activity of Zschemephotocatalyst, even more, a dual Z-scheme photocatalytic scheme has been developed in whichtwo Z-scheme based charge transfer occur simultaneously on a ternary photocatalyst. In this review, wehave discussed the different types of dual Z-schemes along with their application in water purificationand finally, we have discussed the future prospects for further research.
PRBS를 사용하는 Sub-Nyquist 표본화 시스템의 측정행렬 분석
장재혁(Je-Hyuk Jang),최승윤(Seung-Yun Choe),김대영(Dae-Young Kim),Mohamed Yaseen J,Rohit Thakur,이흥노(Heung-No Lee) 대한전자공학회 2015 대한전자공학회 학술대회 Vol.2015 No.6
For a sub-Nyquist sampling system using Pseudo random binary sequences (PRBS), non-ideal characteristics of hardware parts used in the system result in unexpected and unmanageable model mismatches. In this paper, we introduce a representative sub-Nyquist sampling system named Modulated Wide Converter(MWC) and analysis how non-ideal characteristics such as band limited effect and non-linearity influence on a sensing matrix which is essential for recovery. We show that the sensing matrix perturbed by band limited effect can be easily treated and non-linearity case corrupts the output of system as adding error terms which are dependant on the input signal and PRBSs.