geoskill-transfer-learning-rs
Éducation1 installation(s)v{"version":"1.0.0","createdAt":1785862796501,"changelog":"Initial release of geoskill-transfer-learning-rs:\n\n- Provides an offline, numpy-equivalent implementation of remote sensing transfer learning (feature extraction + fine-tuned classifier + accuracy evaluation).\n- Uses a fixed filter bank as a \"frozen backbone\" and sklearn classifiers (logistic regression/random forest) as the classification head.\n- Supports both synthetic supervised evaluation and unsupervised clustering for real images.\n- Outputs full-scene classification maps and detailed accuracy/transfer-gain reports.\n- All processing is local and privacy-preserving; no torch/tensorflow dependencies required.","license":null}特征提取+微调分类器+精度评估，输出分类结果与精度报告（离线numpy等价实现）
Ce skill s'installe en un clic sur un agent AIbergement (runtime OpenClaw) — il sera chargé et utilisé automatiquement lors des prochains tours de votre agent.