In order to meet the urgent needs of refined geological disaster risk assessment at a county scale, and in view of the shortcomings of existing methods in the aspects of sample dependence, rainfall time-varying differences, and vulnerability quantification, this study takes Linxiang City as an example, integrates multi-source data such as geology, geography, meteorology, remote sensing, and field survey, and explores practical methods. A random forest (RF) model was implemented for geological hazard susceptibility mapping, and its hyper-parameters were tuned using Bayesian optimization. Based on a statistical analysis of the frequency of historical disaster events, a risk classification of rainfall in the flood season and non-flood season was evaluated. A vulnerability simplification method based on the value and exposure of disaster-bearing bodies was proposed. Finally, rapid risk assessment was achieved by matrix superposition. The results showed that the model had high accuracy (AUC = 0.903). The use of field survey risk types effectively enhanced the susceptibility sample set and verified the accuracy of risk assessment. The risk factor in the flood season and non-flood season was significantly different, and the very-high- and high-risk areas in the flood season were mainly distributed in the shallow metamorphic rock mountainous area in the east of Yanglousi Town and the granite residual soil area in the south of Zhanqiao Town, the latter of which was highly consistent with the field survey results. This study demonstrated value in terms of sample enhancement, model optimization, consideration of time-varying rainfall, and vulnerability simplification. The evaluation results can provide direct support for the construction of a “point–area dual control” system for geological disasters in Linxiang City, and the methodological framework can also provide a practical reference for risk evaluation in other counties.

Granite residual and slope-wash soils are important disaster-prone geological bodies in the hilly and mountainous areas of Hunan Province, China. Their engineering classification has long relied on manual visual inspection and laboratory testing, which is inefficient and subjective. In this study, an automatic classification method based on deep learning image recognition is proposed for granite residual and slope-wash soils in the mountainous areas of Hunan Province. First, a three-class primary classification scheme was established, comprising residual clay (RNC), residual sandy clay (RNSC), and residual gravelly clay (RNGC), based primarily on the gravel content of particles larger than 2 mm (RNC < 5%, RNSC 5–20%, RNGC > 20%). Second, 7678 geotechnical test records from 21 counties in Hunan Province were collected, and classification labels were assigned through a strategy combining manual verification and automatic inference using Random Forest (5-fold cross-validation macro F1 = 0.913). From approximately 10,096 original field images, 3380 pure soil image patches were retained after segmentation and screening. A dataset of 43,940 samples was then generated through two-stage preprocessing (including denoising and illumination correction) and 13-fold data augmentation. A CNN image classification model was constructed based on a ResNet18 backbone network pre-trained on ImageNet. On the independent test set (6591 images), the primary classification accuracy reached 91.46%, with a macro F1-score of 0.9128; the per-class F1-scores for RNC, RNSC, and RNGC were 0.921, 0.885, and 0.933, respectively. Grad-CAM visualization analysis demonstrated that the model’s attention was primarily focused on soil particle distribution regions rather than non-soil background areas, confirming the effective learning of mixed-grain features. The study shows that the combined application of transfer learning and 13-fold data augmentation can significantly improve classification performance under limited sample size conditions (an improvement of 15.33 percentage points compared to the baseline of 76.13%), demonstrating promising potential for engineering applications.