test_qt_vs_mx

test_qt_vs_mx.py

import cv2
import sys
import numpy as np
import mxnet as mx

from PyQt5.QtWidgets import QApplication

QApplication(sys.argv)  # comment it


def make_input():
    img = cv2.imread('face.jpg')  # aligned
    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    img = np.transpose(img, (2,0,1))
    input_blob = np.expand_dims(img, axis=0)
    data = mx.nd.array(input_blob)
    db = mx.io.DataBatch(data=(data,))
    return db


def fit(model, data):
    print('------- Input (first 10 numbers) -------')
    print(data.data[0][0, 0, 0, :10])
    model.forward(data, is_train=False)
    ret = model.get_outputs()[0].asnumpy()
    print('------- Result (first 10 numbers) -------')
    print(ret[0,0,0,:10])


def get_model():
    sym, arg_params, aux_params = mx.model.load_checkpoint('model/model', 0)
    all_layers = sym.get_internals()
    sym = all_layers['_minusscalar0_output']
    print('------- Output layer info -------')
    print(sym.list_arguments())
    print(sym.list_outputs())
    print(sym.list_attr())
    print(sym.debug_str())
    model = mx.mod.Module(symbol=sym, context=mx.cpu(), label_names = None)
    model.bind(data_shapes=[('data', (1, 3, 112, 112))])
    model.set_params(arg_params, aux_params)
    return model

model = get_model()
db = make_input()
fit(model, db)