Can I rewrite PyTorch YuNet face detection cnn model in GGML?

Can I rewrite a PyTorch [YuNet](https://github.com/ShiqiYu/libfacedetection.train/blob/master/weights/yunet_n.pth) face detection CNN model directly in GGML?
If not, is it possible to convert my PyTorch YuNet (.pth) or ONNX model to GGUF format and run inference with it?