Freezing–thawing action has a great impact on the physical and mechanical deterioration processes of rock materials in cold areas where environmental changes are very complicated. The direct shear test under unloading normal stress was adopted to investigate the shear mechanical behavior of sandstone samples after a freezing–thawing cycle in this paper. The failure shear displacement (D_sf), the failure normal displacement (D_nf), the shear displacement of unloading (D_su), and the normal displacement of unloading (D_nu) were analyzed to describe the evolution of shear and normal deformation during the test. The results indicated that the shear displacement increased as the freezing–thawing cycle duration increased in a direct shear test under unloading normal stress. The unloading rate and the number of freezing–thawing cycles affected the failure pattern of the rock sample significantly in both the direct shear test under unloading normal stress and the direct shear test. The three-dimensional inclination angle, the distortion coefficient, and the roughness correlation coefficient of the fracture surface are dependent on the number of freezing–thawing cycles and the unloading rate. The surface average gradient mode of the fracture surface decreased as the freezing–thawing cycle times and unloading rate rose.

To address the challenges of significant scale variations and difficulty in extracting discriminative features for small targets in airborne remote sensing object detection, while considering the constraints and efficiency requirements of onboard systems, this article proposes a lightweight detection algorithm. First, a structural reparameterization strategy is applied to optimize depthwise separable convolutions, simplifying complex structures during the inference stage, which improves inference speed and memory utilization. Second, the G-Shuffle module is designed to significantly enhance feature extraction efficiency and interchannel information interaction, balancing computational complexity and detection accuracy. Lastly, a dynamic multiscale pyramid network is introduced, employing pyramid convolution to effectively extract and fuse multiscale features. Additionally, a channel attention network is integrated to establish channel interconnections, expanding the receptive field and enhancing the extraction of finegrained features at different scales, thereby improving the detection capability for targets of various sizes, especially small targets. Experimental results demonstrate that the proposed method achieves a mAP@0.5 of 0.552 and mAP@0.5:0.95 of 0.265 on the AI-TOD dataset, with a 2.7 ms improvement in inference speed and a 14% reduction in model parameters. These results indicate superior performance across multiple metrics compared to existing state-of-the-art methods, proving the method's potential and practicality for airborne remote sensing object detection.