为TFliteconverter创建代表性数据集的正确方法是什么？ [英] What is the correct way to create representative dataset for TFliteconverter?

问题描述

我正在尝试使用 INT8 的权重和激活来推断 tinyYOLO-V2 。我可以使用TFliteConverter将权重转换为INT8。对于 INT8 激活，我必须给出代表性的数据集以估计比例因子。我创建此类数据集的方法似乎不对。

I am trying to infer tinyYOLO-V2 with INT8 weights and activation. I can convert the weights to INT8 with TFliteConverter. For INT8 activation, I have to give representative dataset to estimate the scaling factor. My method of creating such dataset seems wrong.

什么是正确的程序？

def rep_data_gen():
    a = []
    for i in range(160):
        inst = anns[i]
        file_name = inst['filename']
        img = cv2.imread(img_dir + file_name)
        img = cv2.resize(img, (NORM_H, NORM_W))
        img = img / 255.0
        img = img.astype('float32')
        a.append(img)
    a = np.array(a)
    print(a.shape) # a is np array of 160 3D images
    img = tf.data.Dataset.from_tensor_slices(a).batch(1)
    for i in img.take(BATCH_SIZE):
        print(i)
        yield [i]
# https://www.tensorflow.org/lite/performance/post_training_quantization
converter = tf.lite.TFLiteConverter.from_keras_model_file("./yolo.h5")
converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.target_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8]
converter.inference_input_type = [tf.int8]
converter.inference_output_type = [tf.int8]
converter.representative_dataset=rep_data_gen
tflite_quant_model = converter.convert()

ValueError：无法设置张量：得到类型为STRING的张量，但预期输入27为FLOAT32类型，名称：input_1

ValueError: Cannot set tensor: Got tensor of type STRING but expected type FLOAT32 for input 27, name: input_1

推荐答案

我使用您的代码读取数据集并发现错误：

I used your code for reading in a dataset and found the error:

img = img .astype（'float32'）应该

img = img.astype('float32') should be

img = img.astype（np.float32）

img = img.astype(np.float32)

希望这有帮助

这篇关于为TFliteconverter创建代表性数据集的正确方法是什么？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！