Yiqi Huang

Dr. Yiqi Huang


Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

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

Representative Publications

View All Publications
24 November 2022 in Agriculture
Verified
Journal Article
Open access
PDF file
YHYiqi Huang
Yiqi Huang
RLRuqi Li
Ruqi Li
XWXiaotong Wei
Xiaotong Wei
ZWZhen Wang
Zhen Wang
TGTianbei Ge
Tianbei Ge
QXQiao Xi
Qiao Xi
Research on the recognition and segmentation of plant diseases in simple environments based on deep learning has achieved relative success. However, under the conditions of a complex environment and a lack of samples, the model has difficulty recognizing disease spots, or its recognition accuracy is too low. This paper is aimed at investigating how to improve the recognition accuracy of the model when the dataset is in a complex environment and lacks samples. First, for the complex environment, this paper uses DeepLabV3+ to segment sugarcane leaves from complex backgrounds; second, focusing on the lack of training images of sugarcane leaves, two data augmentation methods are used in this paper: supervised data augmentation and deep convolutional generative adversarial networks (DCGANs) for data augmentation. MobileNetV3-large, Alexnet, Resnet, and Densenet are trained by comparing the original dataset, original dataset with supervised data augmentation, original dataset with DCGAN augmentation, background-removed dataset, background-removed dataset with supervised data augmentation, and background-removed dataset with DCGAN augmentation. Then, the recognition abilities of the trained models are compared using the same test set. The optimal network selected based on accuracy and training time is MobileNetV3-large. Classification using MobileNetV3-large trained by the original dataset yielded 53.5% accuracy. By removing the background and adding synthetic images produced by the DCGAN, the accuracy increased to 99%.

11 November 2022 in Agronomy
Verified
Journal Article
Open access
PDF file
QXQiao Xi
Qiao Xi
XLXianghuan Liu
Xianghuan Liu
FWFukuan Wang
Fukuan Wang
ZSZhongyu Sun
Zhongyu Sun
Long YangLong Yang
Long Yang
WQWanqiang Qian
Wanqiang Qian
Invasive alien plants (IAPs) are considered to be one of the greatest threats to global biodiversity and ecosystems. Timely and accurate detection technology is needed to identify these invasive plants, helping to mitigate the damage to farmland, fruit trees and woodland. Hyperspectral technology has the potential to identify similar species. However, the challenge remains to simultaneously identify multiple invasive alien plants with similar colors based on image data. The spectral images were collected by a hyperspectral camera with a spectral range of 450–998 nm, and the raw spectra were extracted by Cubert software. First derivative (FD), Savitzky-Golay (SG) smoothing and standard normal variate (SNV) were used to preprocess the raw spectral data, respectively. Then, on the basis of preprocessing, principal component analysis (PCA) and ant colony optimization (ACO) were used for feature dimensionality reduction, and the reduced features were used as input variables for later modeling. Finally, a combination of both dimensionality reduction and non-dimensionality reduction is used for identification using support vector machines (SVM) and random forests (RF). In order to determine the optimal recognition model, a total of 18 combinations of different preprocessing methods, dimensionality reduction methods and classifiers were tested. The results showed that a combination of SG smoothing and SVM achieved a total accuracy (A) of 89.36%, an average accuracy (AA) of 89.39% and an average precision (AP) of 89.54% with a test time of 0.2639 s. In contrast, the combination of SG smoothing, the ACO, and SVM resulted in weaker performance in terms of A (86.76%), AA (86.99%) and AP (87.22%), but with less test time (0.0567 s). The SG-SVM and SG-ACO-SVM models should be selected considering accuracy and time cost, respectively, for recognition of the seven IAPs and background in the wild.

8 February 2026 in Agriculture
Verified
Journal Article
Open access
PDF file
YHYiqi Huang
Yiqi Huang
THTao Huang
Tao Huang
JDJing Du
Jing Du
JQJinxue Qiu
Jinxue Qiu
CLConghui Liu
Conghui Liu
LWLiang Wang
Liang Wang
Ensuring food security and agricultural biosecurity increasingly depends on the rapid and accurate identification of harmful organisms that threaten crop production. Traditional identification methods rely heavily on expert knowledge, are time-consuming, and often fail in complex multi-species scenarios. To address these limitations, this study establishes a comprehensive image dataset that includes three major categories of agricultural harmful organisms—pests, weeds, and crop diseases—and proposes an enhanced convolutional neural network, DenseNet-CSL (DenseNet with Coordinate Attention, Deep Supervision, and Label Smoothing), developed based on DenseNet121 for efficient multi-class recognition. The dataset comprises 62 pest species, 28 weed species, and 30 major crop diseases, totaling 23,995 images collected under diverse growth stages, ecological conditions, and imaging environments. DenseNet-CSL incorporates three targeted improvements: a Coordinate Attention mechanism to strengthen spatial and channel feature representation, Deep Supervision to accelerate convergence and enhance generalization, and Label Smoothing Loss to regularize the output distribution and reduce overconfidence, which is beneficial under imbalanced and noisy data. Experimental results demonstrate that DenseNet-CSL achieves a precision of 81.3%, a recall of 80.1%, and an F1-score of 80% on the constructed dataset—outperforming DenseNet121, ResNet101, EfficientNetV2, and MobileNetV3—while shortening inference time by 1.36 s and adding only 1.772 MB of additional model parameters. These findings highlight the effectiveness of DenseNet-CSL for multi-class recognition of agricultural pests, weeds, and diseases, and underscore the importance of multi-source, multi-scene datasets for improving model robustness and generalization. The proposed framework provides a viable technical pathway for intelligent diagnosis and monitoring of agricultural harmful organisms, supporting port quarantine and agricultural biosecurity applications.

9 February 2024 in Journal of Visualized Experiments
Verified
Journal Article
Closed access
ZHZhenyu Huang
Zhenyu Huang
ZXZhiyong Xu
Zhiyong Xu
YLYanzhou Li
Yanzhou Li
BLBo Liu
Bo Liu
CLConghui Liu
Conghui Liu
YHYiqi Huang
Yiqi Huang
We report on the detailed steps of a method to estimate the biomass of invasive plants based on UAV remote sensing and computer vision. To collect samples from the study area, we prepared a sample square assembly to randomize the sampling points. An unmanned aerial camera system was constructed using a drone and camera to acquire continuous RGB images of the study area through automated navigation. After completing the shooting, the aboveground biomass in the sample frame was collected, and all correspondences were labeled and packaged. The sample data was processed, and the aerial images were segmented into small images of 280 x 280 pixels to create an image dataset. A deep convolutional neural network was used to map the distribution of Mikania micrantha in the study area, and its vegetation index was obtained. The organisms collected were dried, and the dry weight was recorded as the ground truth biomass. The invasive plant biomass regression model was constructed using the K-nearest neighbor regression (KNNR) by extracting the vegetation index from the sample images as an independent variable and integrating it with the ground truth biomass as a dependent variable. The results showed that it was possible to predict the biomass of invasive plants accurately. An accurate spatial distribution map of invasive plant biomass was generated by image traversal, allowing precise identification of high-risk areas affected by invasive plants. In summary, this study demonstrates the potential of combining unmanned aerial vehicle remote sensing with machine learning techniques to estimate invasive plant biomass. It contributes significantly to the research of new technologies and methods for real-time monitoring of invasive plants and provides technical support for intelligent monitoring and hazard assessment at the regional scale.

5 October 2025 in Sensors
Verified
Journal Article
Open access
PDF file
HLHenglong Lin
Henglong Lin
HXHuajie Xue
Huajie Xue
JGJingru Gong
Jingru Gong
HCHuang Cong
Huang Cong
QXQiao Xi
Qiao Xi
YHYiqi Huang
Yiqi Huang
Currently, the customs inspection of wood-boring pests in timber still primarily relies on manual visual inspection, which involves observing insect holes on the timber surface and splitting the timber for confirmation. However, this method has significant drawbacks such as long detection time, high labor cost, and accuracy relying on human experience, making it difficult to meet the practical needs of efficient and intelligent customs quarantine. To address this issue, this paper develops a rapid identification system based on the peristaltic signals of wood-boring pests through the PyQt framework. The system employs a deep learning model with multi-attention mechanisms, namely the Residual Denoising Vision Network (RDVNet). Firstly, a LabVIEW-based hardware–software system is used to collect pest peristaltic signals in an environment free of vibration interference. Subsequently, the original signals are clipped, converted to audio format, and mixed with external noise. Then signal features are extracted through three cepstral feature extraction methods Mel-Frequency Cepstral Coefficients (MFCC), Power-Normalized Cepstral Coefficients (PNCC), and RelAtive SpecTrAl-Perceptual Linear Prediction (RASTA-PLP) and input into the model. In the experimental stage, this paper compares the denoising module of RDVNet (de-RDVNet) with four classic denoising models under five noise intensity conditions. Finally, it evaluates the performance of RDVNet and four other noise reduction classification models in classification tasks. The results show that PNCC has the most comprehensive feature extraction capability. When PNCC is used as the model input, de-RDVNet achieves an average peak signal-to-noise ratio (PSNR) of 29.8 and a Structural Similarity Index Measure (SSIM) of 0.820 in denoising experiments, both being the best among the comparative models. In classification experiments, RDVNet has an average F1 score of 0.878 and an accuracy of 92.8%, demonstrating the most excellent performance. Overall, the application of this system in customs timber quarantine can effectively improve detection efficiency and reduce labor costs and has significant practical value and promotion prospects.