MXNet 多rec参与训练

下面是内容只是简写

import mxnet as mx
import numpy as np
class Mix(mx.io.DataIter):
    def __init__(self, paths, **kwargs):
        super(Mix, self).__init__()
        self.flags = range(len(paths))
        self.numbers = {}.fromkeys(self.flags, 0)
        self.generator = lambda x : mx.io.ImageRecorIter(x, **kwargs)
        self.recs = map(self.generator, paths)

    def reset(self):
        for rec in self.recs:
            rec.reset()
        self.flags = range(len(self.recs))

    def next(self):
        while(self.flags):
            idx = np.random.chioce(self.flags)
            try:
                self.batch = self.recs[idx].next()
            expect StopIteration:
                self.flags.remove(idx)
                if not self.flags:
                    raise StopIteration
                continue
            break
        .......