未捕獲（承諾中）錯誤：未知層

Question

我正在關注google 的圖像分類教程。 在某些時候，完成了一些數據增強： RandomFlip 、 RandomRotation 、 RandomZoom 。

完成培訓后，我使用tfjs.converters.save_keras_model導出了 model 以用於我的節點應用程序。 在節點啟動時，我收到此錯誤：

未捕獲（承諾）錯誤：未知層：RandomFlip。 這可能是由於以下原因之一：該層定義在 Python 中，在這種情況下需要移植到 TensorFlow.js 或您的 Z686155AF75A60A0F6E9D80C1F7EDD3E9 代碼。 自定義層在 JavaScript 中定義，但未使用 tf.serialization.registerClass() 正確注冊。

我在某處讀到我可以以某種方式編輯 model.json 文件以使其工作。 但我不知道如何將其應用於這三個轉換。 這是 modej.json 文件的第一行：

    {"format": "layers-model", "generatedBy": "keras v2.4.0", "convertedBy": "TensorFlow.js Converter v2.4.0", "modelTopology": {"keras_version": "2.4.0", "backend": "tensorflow", "model_config": {"class_name": "Sequential", "config": {"name": "sequential_2", "layers": [{"class_name": "InputLayer", "config": {"batch_input_shape": [null, 200, 200, 3], "dtype": "float32", "sparse": false, "ragged": false, "name": "sequential_1_input"}}, {"class_name": "Sequential", "config": {"name": "sequential_1", "layers": [{"class_name": "InputLayer", "config": {"batch_input_shape": [null, 200, 200, 3], "dtype": "float32", "sparse": false, "ragged": false, "name": "random_flip_input"}}, {"class_name": "RandomFlip", "config": {"name": "random_flip", "trainable": true, "batch_input_shape": [null, 200, 200, 3], "dtype": "float32", "mode": "horizontal", "seed": null}}, {"class_name": "RandomRotation", "config": {"name": "random_rotation", "trainable": true, "dtype": "float32", "factor": 0.1, "fill_mode": "reflect", "interpolation": "bilinear", "seed": null}}, {"class_name": "RandomZoom", "config": {"name": "random_zoom", "trainable": true, "dtype": "float32", "height_factor": 0.1, "width_factor": null, "fill_mode": "reflect", "interpolation": "bilinear", "seed": null}}]}}, {"class_name": "Rescaling", "config": {"name": "rescaling_2", "trainable": true, "dtype": "float32", "scale": 0.00392156862745098, "offset": 0.0}}, {"class_name": "Conv2D", "config": {"name": "conv2d_3", "trainable": true, "dtype": "float32", "filters": 16, "kernel_size": [

有什么辦法可以幫助解決這個問題嗎？

Answer 1

將modelTopology.model_config.class_name從Sequential更改為Model 。

 {"format": "layers-model", "generatedBy": "keras v2.4.0", "convertedBy": "TensorFlow.js Converter v2.4.0", "modelTopology": {"keras_version": "2.4.0", "backend": "tensorflow", "model_config": {"class_name": "Model", "config": {"name": "sequential_2", "layers": [{"class_name": "InputLayer", "config": {"batch_input_shape": [null, 200, 200, 3], "dtype": "float32", "sparse": false, "ragged": false, "name": "sequential_1_input"}}, {"class_name": "Sequential", "config": {"name": "sequential_1", "layers": [{"class_name": "InputLayer", "config": {"batch_input_shape": [null, 200, 200, 3], "dtype": "float32", "sparse": false, "ragged": false, "name": "random_flip_input"}}, {"class_name": "RandomFlip", "config": {"name": "random_flip", "trainable": true, "batch_input_shape": [null, 200, 200, 3], "dtype": "float32", "mode": "horizontal", "seed": null}}, {"class_name": "RandomRotation", "config": {"name": "random_rotation", "trainable": true, "dtype": "float32", "factor": 0.1, "fill_mode": "reflect", "interpolation": "bilinear", "seed": null}}, {"class_name": "RandomZoom", "config": {"name": "random_zoom", "trainable": true, "dtype": "float32", "height_factor": 0.1, "width_factor": null, "fill_mode": "reflect", "interpolation": "bilinear", "seed": null}}]}}, {"class_name": "Rescaling", "config": {"name": "rescaling_2", "trainable": true, "dtype": "float32", "scale": 0.00392156862745098, "offset": 0.0}}, {"class_name": "Conv2D", "config": {"name": "conv2d_3", "trainable": true, "dtype": "float32", "filters": 16, "kernel_size": [