模型仿真测试

1.原始模型文件

采用resnetv1.5预训练模型，可以在github上下载

http://download.tensorflow.org/models/resnet_v1_50_2016_08_28.tar.gz

2.矫正集

链接: https://pan.baidu.com/s/1kNTBEZy3_XdQfS-OSy4fPA 密码: hkkl

3.cfg文件

如附件中resnet_v1.5_run.cfg所示：

[Common]
mode=run

[Parser]
model_name = resnet_v1_50
detection_postprocess = 
model_domain = image_classification
output = resnet_v1_50/predictions/Reshape
input_model = ./model/frozen.pb
input = input
input_shape = [1,224,224,3]
output_dir = ./

[AutoQuantizationTool]
model_name = resnet_v1_50
quantize_method = SYMMETRIC
ops_per_channel = DepthwiseConv
calibration_data = ./dataset/dataset.npy
calibration_label = ./dataset/label.npy
preprocess_mode = normalize
quant_precision=int8
reverse_rgb = False
label_id_offset = 0

[GBuilder]
inputs=./model/input.bin
simulator=aipu_simulator_z1
outputs=output_resnet_v1_50.bin
profile= True
target=Z1_0701

4.simulator执行结果

使用quant_predict.py解析output_v1_50.bin后：

predict first 5 label:
    index  230, prob  65, name: Old English sheepdog, bobtail
    index  231, prob  58, name: Shetland sheepdog, Shetland sheep dog, Shetland
    index  226, prob  35, name: malinois
    index  232, prob  35, name: collie
    index  263, prob  34, name: Brabancon griffon
true first 5 label:
    index  230, prob 109, name: Old English sheepdog, bobtail
    index  231, prob  96, name: Shetland sheepdog, Shetland sheep dog, Shetland
    index  232, prob  57, name: collie
    index  226, prob  54, name: malinois
    index  263, prob  53, name: Brabancon griffon
Detect picture save to result.jpeg

5.附件

链接: https://pan.baidu.com/s/1kNTB... 密码: hkkl