Computer vision-assisted dynamic identification of sealing microfractures in geologic bodies around gas storage wells

The in-depth application of artificial intelligence algorithms in geologic image recognition provides a brand-new path for the automatic detection of fine-grained structures such as microcracks. Aiming at the problems of difficult to recognize microcracks in the hermetic structure of the geological body around the gas storage well and many interference factors, an improved target detection model integrating the bilinear interaction channel attention and the dual-pathway spatial enhancement mechanism is constructed in the study. The model is based on YOLOv5, and integrates the channel dimension and spatial dimension saliency enhancement modules to improve the extraction and expression ability of low-contrast, fine linear crack features. The experimental results show that the proposed model achieves 0.931, 0.918, 0.894 and 0.906 in terms of accuracy, precision, recall and F1 value, respectively, which are significantly better than the comparison model; at the same time, the root-mean-square error is reduced to 0.109, and the average inference time is shortened to 0.28 seconds, which shows excellent recognition accuracy and response efficiency. In the multi-type image test, the model still maintains the recognition accuracy of not less than 0.88 in complex scenes, showing strong generalization ability. The results verify the synergistic enhancement of the channel and spatial attention mechanism, which provides an effective solution for intelligent crack detection in complex geological environments.