![](/img/trans.png)
[英]How to convert an object detection model, in it's frozen graph, to a .tflite, without any knowledge of input and output arrays
[英]How to set input and output arrays while converting to a tflite model
import tensorflow as tf
from tensorflow.contrib import lite
graph_def_file = 'D:\\Models\\kapl\\inference_graph \\frozen_inference_graph.pb'
input_arrays = [1,600,1024,1] #image_tensor
output_arrays = [1,600,1024,1]
converter = tf.contrib.lite.TFLiteConverter.from_frozen_graph(graph_def_file, input_arrays, output_arrays)
converter.post_training_quantize = True
tflite_quantized_model = converter.convert()
open("quantized_model.tflite", "wb").write(tflite_quantized_model)
input_arrays
和output_arrays
表示模型图的输入和输出张量。
检查从图表的最简单方式.pb
文件是使用summarize_graph工具。
如果该方法产生错误,则可以使用TensorBoard可视化GraphDef,并在图形中查找输入和输出。 要可视化.pb
文件,请使用import_pb_to_tensorboard.py
脚本,如下所示:
python import_pb_to_tensorboard.py --model_dir <model path> --log_dir <log dir path>
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.