geo-deep-learning

Invoke before recommending, training, or auditing a neural method for geospatial imagery, including vision transformers, U-Net/DeepLab/SegFormer, object detection, pixel classification, building/road extraction, and EO foundation-model fine-tuning. Also invoke for neural chip-split validity, IoU/accuracy claims, augmentation, imbalanced losses, spatial validation, or sliding-window inference. Use…