Hung Quang Ha

Mr. Hung Quang Ha

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

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

Representative Publications

View All Publications
1 July 2025 in Indonesian Journal of Electrical Engineering and Computer Science
Verified
Journal Article
Open access
PDF file
HHHung Quang Ha
Hung Quang Ha
THTrong-Minh Hoang
Trong-Minh Hoang
MPMinh Trien Pham
Minh Trien Pham
Plant diseases significantly impact the quality and productivity of crops, leading to substantial economic losses. This paper introduces two enhanced EfficientNet-B5 architectures, EfficientNetB5-sigca and EfficientNetB5- sigbi, specifically designed to detect and classify diseases in cucurbit leaves. We employ EfficientNet-B5 for feature extraction, using a 456×456×3 input and omitting the top layer to generate feature maps with Swish activation. A global average pooling 2D layer replaces the conventional fully connected layer, producing a flattened vector. This is followed by a dense layer with four output units, L2 regularization, and sigmoid activation, using either categorical or binary cross-entropy as the loss function. We also developed a novel image dataset targeting cucumber and cantaloupe leaves, including 11,425 augmented images categorized into four disease classes: anthracnose, powdery mildew, downy mildew, and fresh leaf. Our experiments dataset demonstrates that the EfficientNetB5-sigbi achieves an accuracy of 97.07%, marking a significant improvement in classifying similar diseases in cucurbit leaves.

6 October 2026 in Agriculture
Verified
Journal Article
Open access
PDF file
HHHung Quang Ha
Hung Quang Ha
HCHa Duc Chu
Ha Duc Chu
NCNguyễn Nguyên Chương
Nguyễn Nguyên Chương
TPTruong Trinh Phung
Truong Trinh Phung
TATouhidur Rahman Anik
Touhidur Rahman Anik
Lam-Son Phan TranLam-Son Phan Tran
Lam-Son Phan Tran
Fruit maturity and ripeness determine harvest timing, postharvest management, market quality, and consumer acceptance. Conventional assessments commonly depend on destructive physicochemical analyses or subjective visual inspection, which limits rapid and repeated evaluation across production and supply chains. This review synthesizes recent advances in non-destructive fruit maturity and ripeness assessment using machine vision, visible and near-infrared spectroscopy, hyperspectral and multispectral imaging, acoustic and mechanical techniques, electronic noses, and dielectric measurements. Particular attention is given to data preprocessing, feature extraction, chemometric modeling, conventional machine learning, and deep learning approaches used to associate sensor responses with firmness, soluble solids content, acidity, pigment composition, internal defects, and physiological stage. The reviewed evidence indicates that spectroscopy and imaging provide strong predictive performance, while multimodal sensing can improve reliability by combining complementary external and internal attributes. However, differences in cultivar, orchard conditions, instruments, sampling protocols, reference measurements, and validation strategies restrict model transferability and cross-study comparison. The review further identifies the gaps that most constrain deployment, which include the absence of standardized maturity definitions, the scarcity of multisite datasets with external validation, the limited reporting of explainable models and uncertainty estimates, and the few sensor-fusion systems tested under edge-deployment conditions.

9 July 2026 in CTU Journal of Innovation and Sustainable Development
Verified
Journal Article
Open access
PDF file
QVQuyet Nguyen Van
Quyet Nguyen Van
HHHung Quang Ha
Hung Quang Ha
TTTrinh Phung Truong
Trinh Phung Truong
TMTrung Vu Minh
Trung Vu Minh
HCHa Duc Chu
Ha Duc Chu
TMTrien Pham Minh
Trien Pham Minh
Durian (Durio zibethinus) is a high-value fruit crop in Viet Nam, with rapidly increasing production, which underscores the need for accurate leaf condition recognition in real orchards. Field images are challenging because symptoms vary in scale and are captured under uncontrolled illumination and complex backgrounds, while many deep models improve accuracy at the cost of higher computation. This study proposes an efficient YOLOv8-based classifier for durian leaf images. The method removes the deepest downsampling stage to preserve higher-resolution features and introduces PartialTriDW, a lightweight tri-branch depthwise block with adaptive softmax fusion and low-cost channel mixing, to enhance multi-scale texture representation in the early and mid-backbone stages. Experiments on a five-class dataset of 4,437 field images, trained from scratch, show that the YOLOv8 baseline achieves 0.918 accuracy, the CutP5 variant achieves 0.927 accuracy, and the proposed model reaches 0.949 accuracy with 0.56 million parameters and 3.3 GFLOPs, while maintaining comparable latency. The results indicate that preserving spatial detail and strengthening multi-scale texture modeling improve practical durian leaf classification with controlled computational cost.

1 April 2026
Verified
Book Chapter
Closed access
VNVan Quyet Nguyen
Van Quyet Nguyen
HHHung Quang Ha
Hung Quang Ha
QVQuang Huy Vuong
Quang Huy Vuong
MVMinh-Trung Vu
Minh-Trung Vu
THTrong-Minh Hoang
Trong-Minh Hoang
MPMinh Trien Pham
Minh Trien Pham

1 January 2025 in Applied Engineering in Agriculture
Verified
Journal Article
Closed access
HHHung Quang Ha
Hung Quang Ha
TDTri Dung Do
Tri Dung Do
TTThanh-Huong Tran
Thanh-Huong Tran
THTrong-Minh Hoang
Trong-Minh Hoang
TTTruc Mai Tran
Truc Mai Tran
MPMinh Trien Pham
Minh Trien Pham
Highlights The hybrid CNN-BiLSTM model accurately predicts 60-day air temperatures for FAW life cycle analysis. Temperature-based algorithm forecasts FAW stages, enabling proactive pest management strategies. CNN-BiLSTM outperforms standalone models, achieving an R2 of 0.90 with minimal prediction error. Accurate life stage predictions support timely pesticide applications and FAW population control. Abstract. The fall armyworm poses a threat to global food security by destroying major crops, including maize. Its rapid lifecycle, high mobility, and adaptability to diverse climates make early detection and prediction of its development stages crucial for timely pest control interventions. However, existing studies rarely combine model temperature forecasting with biological phase estimation, and few are tailored for regional datasets in tropical agricultural zones, such as Vietnam. This study introduces a novel life-cycle prediction algorithm that integrates deep-learning-based 60-day temperature forecasts with a temperature-phase mapping strategy. Three models—Long Short-Term Memory, Convolutional Neural Network, and a hybrid CNN-BiLSTM—were evaluated and trained on 10 years of meteorological data from Vinh Phuc Province, Vietnam. The hybrid model achieved the best performance, with a Mean Absolute Error of 1.21°C, a Root Mean Squared Error of 1.57, and an R 2 value of 0.90. The predicted temperatures were used to estimate FAW life stages, and the estimated adult phase period showed high consistency with the observed timing of adult FAW population peaks in the field. This work highlights the potential of combining weather forecasting and biological modelling to enhance early warning systems for pest management in vulnerable agricultural regions. Keywords: Bidirectional long short-term memory, Convolutional neural network, Fall armyworm, Life cycle prediction, Long short-term memory.

14 October 2026 in Tạp chí Khoa học và Công nghệ Trường Đại học Công nghệ Đông Á
Verified
Journal Article
Closed access
HLHoàng Vũ Lê
Hoàng Vũ Lê
HTHà Phương Trần
Hà Phương Trần
TTThị Thu Phương Trần
Thị Thu Phương Trần
HHHung Quang Ha
Hung Quang Ha
MPMinh Triển Phạm
Minh Triển Phạm
MVMinh Trung Vũ
Minh Trung Vũ
The sweet potato weevil (Cylas formicarius) is a common pest insect that causes significant economic losses by reducing both yield and tuber quality. This study develops a recognition pipeline for detecting small, densely distributed objects, aiming to provide an early pest forecasting tool and optimize the use of plant protection chemicals. The dataset consists of 334 field images with densities ranging from 10 to 100 individuals per image, annotated in YOLO format. Experimental comparisons of performance among YOLOv8, YOLOv9, YOLOv10, and YOLOv11 were conducted under the same data conditions and training configurations. The results indicate that YOLOv11n achieves the best performance, with a Precision of 99.12%, Recall of 97.78%, mAP@0.5 of 98.77%, and mAP@[0.5:0.95] of 77.15%. With an optimized architecture comprising 2.65 million parameters and 6.6 GFLOPs, this model ensures compactness for deployment. A software system was developed to integrate the model, visualize the results, and validate its practical applicability in agricultural pest monitoring.