量化目标检测模型

Question

[2]  frozen_graph_file = # path to frozen graph (.pb file)
[3]  input_arrays = ["normalized_input_image_tensor"]
[4]  output_arrays = ['TFLite_Detection_PostProcess',
[5]            'TFLite_Detection_PostProcess:1',
[6]            'TFLite_Detection_PostProcess:2',
[7]            'TFLite_Detection_PostProcess:3']
[8]  input_shapes = {"normalized_input_image_tensor" : [1, 300, 300, 3]}
[9]
[10] converter = tf.lite.TFLiteConverter.from_frozen_graph(frozen_graph_file,
[11]                                                  input_arrays=input_arrays,
[12]                                                  output_arrays=output_arrays,
[13]                                                  input_shapes=input_shapes)
[14] converter.allow_custom_ops = True
[15] converter.optimizations = [tf.lite.Optimize.DEFAULT]
[16] tflite_quant_model = converter.convert()
[17] with open(tflite_model_quant_file, "wb") as tflite_file:
[18]     tflite_file.write(tflite_model_quant)

在量化模型时，我们通常为模型提供一些校准数据以识别激活范围，从而定义尺度和零点。这是为张量量化完成的。物体检测边界框坐标的量化值是如何获得的？它是否遵循相同的时尚？ 在 Tensorflow 中，它们为无法以常规方式量化的操作提供自定义操作。我在哪里可以获得它们的详细实现，尤其是 TFLite_Detection_PostProcess？

Answer 1

TFLite_Detection_PostProcess 的实现在https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/kernels/detection_postprocess.cc

不管量化如何，TFLite_Detection_PostProcess 的输出总是float，所以我认为你不需要在这里关心它。