Novel approach for soil classification using machine learning methods

Nguyen, Manh Duc; Costache, Romulus; Sy, An Ho; Ahmadzadeh, Hassan; Le, Hiep Van; Prakash, Indra; Pham, Binh Thai

doi:10.1007/s10064-022-02967-7

“…Models Used Accuracy Obtained Manh Duc Nguyen et al [5] SVC, MLP, RF SVC=98.4% Kumar et al [30] ANN 95% Choudhury et. al [31] SVM 91.37%, 95.72% Srunitha et al [32] SVM 95% Lu et al [33] CNN AUC=91.47% Lamesck et al [34] SVM-poly 94.3% Bhattacharya et al [35] Decision Trees, ANN and SVM 89.34%, 87% and 71.18% Zhao et al [36] ANN 88%, 81% Mengistu et al [37] Back-Propagation Neural Network (BPNN) 89.7% Wu et al [38] Multi SVM with Polynomial Kernel 79.4% and 99.2% Guang et al [39] PLS-DA and Multi SVM with Polynomial Kernel 93.33% and 96.67% Vibhute et al [40] Multi…”

Section: Referencesmentioning

confidence: 99%

“…In 2022, Nguyen et al [5] proposed a novel classification approach for determining different soil classes utilizing machine learning approaches, specifically support vector classification (SVC), random forest (RF), and multilayer perceptron (MLP) models. The models were developed using a database of 4888 soil samples obtained from Vietnam projects, with 15 soil properties factors selected as input parameters to categorize the samples into 5 soil classes.…”

Section: Introductionmentioning

confidence: 99%

Advancing Soil Image Classification through Comprehensive Evaluation for Crop Suggestion

R¹,

Ashoka²,

V

³

2023

Preprint

0

View full text Add to dashboard Cite

Soil image classification is an important task in agricultural and environmental applications. With the advent of deep learning, image-based soil classification has become a popular research topic. Transfer learning is a widely used approach in image classification, where a pre-trained model is fine-tuned on a new dataset. However, traditional transfer learning approaches only fine-tune the last layer of the pre-trained model, which may not be sufficient for achieving high performance on a new task. In this paper, we propose HybridTransferNet, a novel hybrid transfer learning approach for image-based soil classification. HybridTransferNet fine-tunes a small number of earlier layers in addition to the last layer of a pre-trained ResNet50 model, resulting in improved classification performance compared to traditional transfer learning approaches. We evaluate HybridTransferNet on a soil classification dataset and report multiple performance metrics, including accuracy, precision, recall, and F1 score. Our experimental results show that HybridTransferNet outperforms traditional transfer learning approaches and achieves state-of-the-art performance on the soil classification task.

show abstract

“…In 2022, Nguyen et al (5) proposed a new classification method for determining different soil classes using machine learning approaches, specifically support vector classification (SVC), multilayer perceptron (MLP), and random forest (RF) models. The models were developed using a database of 4888 soil samples obtained from Vietnam projects, with 15 soil properties factors selected as input parameters to classify the samples into 5 soil classes.…”

Section: Summary Of Previous Workmentioning

confidence: 99%

“…In recent years, image-based soil classification (5) (6) has become a popular research topic due to advances in imaging technology and deep learning techniques. Deep learning models (7) can extract complex features from soil images and accurately classify them into different categories.…”

Section: Introductionmentioning

confidence: 99%

HybridTransferNet: Advancing Soil Image Classification through Comprehensive Evaluation of Hybrid Transfer Learning

R¹,

Ashoka²,

V

³

2023

Preprint

0

View full text Add to dashboard Cite

Soil image classification is an important task in agricultural and environmental applications. With the advent of deep learning, image-based soil classification has become a popular research topic. Transfer learning is a widely used approach in image classification, where a pre-trained model is fine-tuned on a new dataset. However, traditional transfer learning approaches only fine-tune the last layer of the pre-trained model, which may not be sufficient for achieving high performance on a new task. In this paper, we propose HybridTransferNet, a novel hybrid transfer learning approach for image-based soil classification. HybridTransferNet fine-tunes a small number of earlier layers in addition to the last layer of a pre-trained ResNet50 model, resulting in improved classification performance compared to traditional transfer learning approaches. We evaluate HybridTransferNet on a soil classification dataset and report multiple performance metrics, including accuracy, precision, recall, and F1 score. Our experimental results show that HybridTransferNet outperforms traditional transfer learning approaches and achieves state-of-the-art performance on the soil classification task.

show abstract

“…Rauter and Tschuchnigg (2021) developed a classification model based on a dataset, and the algorithms used were SVM, ANN, and Randomfrest with different combinations of input variables. Nguyen (2022) proposes a new classification method for soil samples using three machinelearning models (SVC, MLP, and RF) and 15 input parameters. A database of 4888 soil samples from Vietnam was used for model development.…”

Section: Introductionmentioning

confidence: 99%

An integrated machine learning model for soil category classification based on CPT

Bai,

Shen,

Zhang

2023

Preprint

0

View full text Add to dashboard Cite

Soil classification is a critical issue in geological engineering, with the Cone Penetration Test (CPT) being an effective in-situ testing technique to record the variation of soil characteristics. Despite many studies that have been conducted on the relationship between CPT parameters and soil categories, analyzing soil in specific areas is essential due to the high uncertainty of geotechnical. In this study, we analyzed CPT parameters and soil categories based on geological soil layers in the Shanghai region. The CPT-based indirect method requires additional geotechnical parameters, which are limited due to the lack of advanced equipment to measure pore pressure in China. To satisfy practical application requirements, we proposed a new CPT-based direct method based on an integrated machine-learning model. By establishing multiple classification models and using Particle Swarm Optimization (PSO) to determine each model's weights, the results of multiple models were integrated to improve classification accuracy and robustness. The experimental results show that the integrated model has high accuracy and robustness in multiple engineering sites, with significant advantages over conventional CPT-based direct methods.

show abstract

Novel approach for soil classification using machine learning methods

Cited by 21 publications

References 36 publications

Advancing Soil Image Classification through Comprehensive Evaluation for Crop Suggestion

Advancing Soil Image Classification through Comprehensive Evaluation for Crop Suggestion

HybridTransferNet: Advancing Soil Image Classification through Comprehensive Evaluation of Hybrid Transfer Learning

An integrated machine learning model for soil category classification based on CPT

Contact Info

Product

Resources

About