Lei Huang

Lei Huang

Auto-generated profile, based on publicly available information

Lei Huang

Yangtze Delta Region Institute (Huzhou), University of Electronic Science and Technology of China, Huzhou, China

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2021 in Talanta
Verified
Journal Article
Closed access
LHLei Huang
Lei Huang
STShulin Tian
Shulin Tian
WZWenhao Zhao
Wenhao Zhao
KLKe Liu
Ke Liu
JGJinhong Guo
Jinhong Guo

27 June 2024 in IEEE Sensors Journal
Verified
Journal Article
Closed access
LHLei Huang
Lei Huang
ZDZhuoyue Ding
Zhuoyue Ding
CZCheng Zhang
Cheng Zhang
RYRun Ye
Run Ye
BYBin Yan
Bin Yan
JGJinhong Guo
Jinhong Guo
Forest fires can cause tremendous damage to the environment and irreparable loss to individual properties. Using deep learning algorithms on embedded sensing devices to detect smoke and flame for real-time detection of forest fire is essential to minimize the risk to the environment and individual properties. This paper proposes an ultra-lightweight network, YOLO-ULNet, for real-time detection of forest fire on embedded sensing devices. Initially, a lightweight network, YOLO-LNet, consisting of input, backbone, neck, and output head was designed to reduce the parameter quantity and computation quantity. The validation results of YOLO-LNet on the computer showed that the mAP@0.5, speed, computation quantity, and parameter quantity were 71.2%, 110 FPS, 0.7 GFLOPs, 0.26 M, respectively, representing a performance that was superior to that of YOLOv5s. The test results of YOLO-LNet on the Raspberry Pi (RPi) 4B showed that the precision and speed were 77.75% and 14.39 FPS, respectively, where the speed required for real-time detection (20 FPS) still could not be achieved. Subsequently, channel pruning and feature distillation model compression methods were employed to obtain the ultra-lightweight network YOLO-ULNet with improved speed and precision. The validation results of YOLO-ULNet on the computer showed that the mAP@0.5, speed, computation quantity, and parameter quantity were 69.1%, 115 FPS, 0.4 GFLOPs, and 0.19 M, respectively. Meanwhile, the test results of YOLO-ULNet on the RPi 4B showed that the precision and speed were 74.50% and 24.57 FPS, respectively, meeting the requirements of the real-time detection of forest fire.

27 January 2022 in IEEE Transactions on Nanobioscience
Verified
Journal Article
Closed access
LHLei Huang
Lei Huang
STShulin Tian
Shulin Tian
WZWenhao Zhao
Wenhao Zhao
KLKe Liu
Ke Liu
Xing MaXing Ma
Xing Ma
JGJinhong Guo
Jinhong Guo
Methamphetamine is a powerful stimulant drug, the abuse of which threatens human health and social stability. Rapid and accurate quantification of methamphetamine is essential to inhibit the abuse and prevalence of methamphetamine effectively. In this paper, we present a portable fluorescence reader with upconverting nanoparticle-labeled lateral flow immunoassay (LFIA) for rapid and accurate quantification of methamphetamine. Based on specific binding of a methamphetamine antigen to an antibody in the LFIA, the fluorescence reader is designed to capture and record the fluorescence intensities T and C of the test and control lines, respectively, and the T/C ratio is calculated to determine the concentration of methamphetamine. The linear range for methamphetamine is 0.1–100 ng/mL. Because the sensor is often susceptible to noise interference, using only the T/C ratio to distinguish weakly positive and negative samples of methamphetamine renders the results inaccurate. To solve this problem, we applied a convolutional neural network (CNN) to learn image features of different methamphetamine concentrations (0, 0.01, 0.05, 0.1, and 0.5 ng/mL) for accurate detection of weakly positive and negative samples. The results show that the proposed method can effectively detect weakly positive and negative samples of methamphetamine with an accuracy of up to 92%. The CNN provides a novel scheme for accurate analysis of weakly positive and negative samples in upconverting nanoparticle-labeled LFIA.

1 November 2018 in Review of Scientific Instruments
Verified
Journal Article
Closed access
PDF file
LHLei Huang
Lei Huang
STShulin Tian
Shulin Tian
KLKe Liu
Ke Liu
GGGuangkun Guo
Guangkun Guo
Yindong XiaoYindong Xiao
Yindong Xiao
XYXing Yang
Xing Yang
The Time Marker Generator (TMG) is an important instrument that is used to calibrate the time base of an oscilloscope. If a direct digital frequency synthesizer is used to generate the necessary waveforms, there is serious distortion in the generated square wave when the ratio of the sampling frequency to the output frequency is non-integer. In addition, the look-up table (LUT) in the direct digital waveform synthesizer needs a large storage capacity to generate a narrow triangular wave at a low output frequency. This paper proposes a design that will generate the time marker whose samples are synthesized by real-time calculation instead of storing them in a LUT. We also propose an M waveform data synthesizer with a parallel structure (M-WDSPS) to reduce the operating clock frequency of the field programmable gate array (FPGA). We built a 4-WDSPS TMG based on this design and reduced the required clock frequency from 1 GHz to 250 MHz, thus making the implementation possible in an FPGA. Our proposed TMG design can provide square wave, pulse wave, narrow triangular wave, and linear triangular wave with different amplitudes in two operating modes: normal mode and highlight mode. The TMG we constructed can provide the time marker with an output frequency range from 10 mHz to 111 MHz, a rising/falling edge time lower than 1 ns, regardless of whether the output frequency is low or high, and a frequency stability lower than 0.1 ppm.

14 June 2019 in IEEE Access
Verified
Journal Article
Open access
YXYindong Xiao
Yindong Xiao
YCYu Chen
Yu Chen
KLKe Liu
Ke Liu
LHLei Huang
Lei Huang
XYXing Yang
Xing Yang
The arbitrary waveform generator is now commonly used to generate waveforms in many fields. The Nyquist sampling theorem only provides an open interval for the sampling rate selection, but how to determine a specific sampling rate within this open interval has not been discussed. In practice, the sample rate conversion is adopted to convert the arbitrarily selected sample rate to a safe one which keeps the mirror frequency out of a fixed passband of an anti-aliasing filter in an arbitrary waveform generator whose sample rate is variable. To select a safe sample rate with low computational cost and memory requirements for conversion, an algorithm, with fractional sample rate conversion in mind, is proposed in this paper to guarantee the safety of the target sample rate and the bound of resource cost. The experimental result shows that not only the computational cost and memory requirements of sampling rate conversion but also the frequency stability and harmonics of the generated signal in the proposed algorithm overcomes the method that arbitrary converts the sampling rate to a fixed maximum one.

1 August 2019
Verified
Conference Paper
Closed access
LYLinglong Yin
Linglong Yin
STShulin Tian
Shulin Tian
KLKe Liu
Ke Liu
Yindong XiaoYindong Xiao
Yindong Xiao
LHLei Huang
Lei Huang
WZWenhao Zhao
Wenhao Zhao
Hybrid Filter Bank digital-to-analog converters suffer from the mismatch among channels. In this paper, a discrete-time behavioral model is derived to describe the effect of nonideality in Hybrid Filter Bank digital-to-analog converters (DAC). Mismatches including nonideal mixers and imperfect anti-aliasing are considered in this model. Based on this model, an integrated digital pre-compensation method is proposed to mitigate the performance degradation. Simulation results shows the validity of the compensation method.