Mapping landslide cracks is essential in understanding landslide dynamics and evolution across space and time. Each landslide crack’s depth, direction, and width are related to the stress and strain imposed on the landslide body. Moreover, their spatial distribution can indicate areas where the landslide can extend, mainly if located in the upper part of the main landslide scarp. Even though the cracks leave a distinct pattern on the landslide body when fresh or when there is a high contrast between the bare soil and surrounding vegetation, these patterns gradually diminish over time, making their detection difficult. The problem of landslide cracks mapping in various environmental conditions and having different ages was tackled in the current study using very high spatial resolution unmanned aerial vehicle (UAV) aerial imagery and derived products in conjunction with deep learning models. U-Net and DeepLab CNN models were applied using masked and non-masked training samples with different tile sizes. As the tile size decreases, the performance metrics, such as precision, recall, and F1-score, generally decrease. Overall, the lowest accuracy was approximately 0.79 for non-mask samples and tile size of 64 pixels, and reached over 0.93 for masked samples and tile size of 512 pixels.