juliensimon

import mxnet as mx
import numpy as np
import logging

logging.basicConfig(level=logging.INFO)

sample_count = 1000
train_count = 800
valid_count = sample_count - train_count

juliensimon

import numpy as np
import cv2

def splitRGBImage(filename):
        img = cv2.imread(filename)
        red = np.copy(img)
        red[:,:,0].fill(0)
        red[:,:,1].fill(0)
        red = cv2.resize(red, (224, 224,))
        cv2.imwrite("red_"+filename, red)

juliensimon

import mxnet as mx
import numpy as np
import cv2
from collections import namedtuple

def loadInceptionv3():
        sym, arg_params, aux_params = mx.model.load_checkpoint('Inception-BN', 0)
        mod = mx.mod.Module(symbol=sym)
        mod.bind(for_training=False, data_shapes=[('data', (1,3,224,224))])
        mod.set_params(arg_params, aux_params)

juliensimon

import mxnet as mx
import numpy as np
import cv2,sys,time
from collections import namedtuple

def loadModel(modelname):
        t1 = time.time()
        sym, arg_params, aux_params = mx.model.load_checkpoint(modelname, 0)
        t2 = time.time()
        t = 1000*(t2-t1)

juliensimon

import struct
import numpy as np
import cv2

labelfile = open("train-labels-idx1-ubyte")
# Read packed structure - big-endian, 2 integers: a magic number and the number of labels
magic, num = struct.unpack(">II", labelfile.read(8))
labelarray = np.fromstring(labelfile.read(), dtype=np.int8)
print labelarray.shape
# Print first labels

juliensimon

import mxnet as mx
import numpy as np
import cPickle
import cv2

def extractImagesAndLabels(path, file):
    f = open(path+file, 'rb')
    dict = cPickle.load(f)
    images = dict['data']
    images = np.reshape(images, (10000, 3, 32, 32))

juliensimon

import mxnet as mx

def buildTrainingSet(path):
    training_data = []
    training_label = []
    for f in ("data_batch_1", "data_batch_2", "data_batch_3", "data_batch_4", "data_batch_5"):
        imgarray, lblarray = extractImagesAndLabels(path, f)
        if not training_data:
            training_data = imgarray
            training_label = lblarray

juliensimon

import mxnet as mx

train_iter = mx.io.ImageRecordIter(
  path_imgrec="cifar10_train.rec", data_name="data", label_name="softmax_label", 
  batch_size=batch, data_shape=(3,28,28), shuffle=True)

valid_iter = mx.io.ImageRecordIter(
  path_imgrec="cifar10_val.rec", data_name="data", label_name="softmax_label", 
  batch_size=batch, data_shape=(3,28,28))

juliensimon

import mxnet as mx

epochs = 100

sym, arg_params, aux_params = mx.model.load_checkpoint("resnext-101",0)
mod = mx.mod.Module(symbol=sym, context=(mx.gpu(0), mx.gpu(1), mx.gpu(2), mx.gpu(3)))
mod.bind(data_shapes=train_iter.provide_data, label_shapes=train_iter.provide_label)
mod.set_params(arg_params, aux_params)
mod.fit(train_iter, eval_data=valid_iter, 
        optimizer_params={'learning_rate':0.05, 'momentum':0.9}, num_epoch=epochs)

juliensimon

import mxnet as mx
import numpy as np
import cv2, cPickle, logging
from symbols import resnext

logging.basicConfig(level=logging.DEBUG)

path="cifar-10-batches-py/"
examples=50000
batch=128