This is an open-source face detection solution built on the Caffe framework. It achieves network lightweighting by removing fully connected layers, supports multi-GPU parallel training, and offers a complete toolchain spanning data preprocessing, model training, and performance evaluation.