Created by: hacksman

代码如下所示：

from PIL import Image
import numpy as np
import paddle.v2 as paddle

# 设置是否用gpu，0为否，1为是
with_gpu = os.getenv('WITH_GPU', '0') != '0'

# 定义网络结构
def convolutional_neural_network_org(img):
    # 第一层卷积层
    conv_pool_1 = paddle.networks.simple_img_conv_pool(
        input=img,
        filter_size=5,
        num_filters=20,
        num_channel=1,
        pool_size=2,
        pool_stride=2,
        act=paddle.activation.Relu())
    # 第二层卷积层
    conv_pool_2 = paddle.networks.simple_img_conv_pool(
        input=conv_pool_1,
        filter_size=5,
        num_filters=50,
        num_channel=20,
        pool_size=2,
        pool_stride=2,
        act=paddle.activation.Relu())
    # 全连接层
    predict = paddle.layer.fc(
        input=conv_pool_2, size=10, act=paddle.activation.Softmax())
    return predict

def main():
    # 初始化定义跑模型的设备
    paddle.init(use_gpu=with_gpu, trainer_count=1)

    # 读取数据
    images = paddle.layer.data(
        name='pixel', type=paddle.data_type.dense_vector(784))
    label = paddle.layer.data(
        name='label', type=paddle.data_type.integer_value(10))

    # 调用之前定义的网络结构
    predict = convolutional_neural_network_org(images)

    # 定义损失函数
    cost = paddle.layer.classification_cost(input=predict, label=label)

    # 指定训练相关的参数
    parameters = paddle.parameters.create(cost)

    # 定义训练方法
    optimizer = paddle.optimizer.Momentum(
        learning_rate=0.1 / 128.0,
        momentum=0.9,
        regularization=paddle.optimizer.L2Regularization(rate=0.0005 * 128))

    # 训练模型
    trainer = paddle.trainer.SGD(
        cost=cost, parameters=parameters, update_equation=optimizer)


    lists = []

    # 定义event_handler，输出训练过程中的结果
    def event_handler(event):
        if isinstance(event, paddle.event.EndIteration):
            if event.batch_id % 100 == 0:
                print "Pass %d, Batch %d, Cost %f, %s" % (
                    event.pass_id, event.batch_id, event.cost, event.metrics)
        if isinstance(event, paddle.event.EndPass):
            # 保存参数
            with open('params_pass_%d.tar' % event.pass_id, 'w') as f:
                parameters.to_tar(f)

            result = trainer.test(reader=paddle.batch(
                paddle.dataset.mnist.test(), batch_size=128))
            print "Test with Pass %d, Cost %f, %s\n" % (
                event.pass_id, result.cost, result.metrics)
            lists.append((event.pass_id, result.cost,
                          result.metrics['classification_error_evaluator']))

    trainer.train(
        reader=paddle.batch(
            paddle.reader.shuffle(paddle.dataset.mnist.train(), buf_size=8192),
            batch_size=128),
        event_handler=event_handler,
        num_passes=10)

    # 找到训练误差最小的一次结果
    best = sorted(lists, key=lambda list: float(list[1]))[0]
    print 'Best pass is %s, testing Avgcost is %s' % (best[0], best[1])
    print 'The classification accuracy is %.2f%%' % (100 - float(best[2]) * 100)

    # 加载数据
    def load_image(file):
        im = Image.open(file).convert('L')
        im = im.resize((28, 28), Image.ANTIALIAS)
        im = np.array(im).astype(np.float32).flatten()
        print im
        im = im / 255.0 * 2.0 - 1.0
        print '-----'
        print im
        return im

    # 测试结果
    test_data = []
    cur_dir = os.path.dirname(os.path.realpath(__file__))
    test_data.append((load_image(cur_dir + '/image/infer_7.png'), ))

    probs = paddle.infer(
        output_layer=predict, parameters=parameters, input=test_data)
    lab = np.argsort(-probs)  # probs and lab are the results of one batch data
    print lab
    print "Label of image/infer_7.png is: %d" % lab[0][0]

if __name__ == '__main__':

    from datetime import datetime
    start_time = datetime.now()
    main()
    end_time = datetime.now()
    escape = end_time-start_time

    print escape

返回的结果如下：

Pass 0, Batch 0, Cost 3.637740, {'classification_error_evaluator': 0.875}
Pass 0, Batch 100, Cost 0.666954, {'classification_error_evaluator': 0.234375}
Pass 0, Batch 200, Cost 0.720760, {'classification_error_evaluator': 0.2265625}
Pass 0, Batch 300, Cost 0.423178, {'classification_error_evaluator': 0.125}
Pass 0, Batch 400, Cost 0.310293, {'classification_error_evaluator': 0.0625}
Test with Pass 0, Cost 0.278447, {'classification_error_evaluator': 0.0803999975323677}
# 为避免太长，中间的部分输出省略掉
Pass 9, Batch 0, Cost 0.069227, {'classification_error_evaluator': 0.015625}
Pass 9, Batch 100, Cost 0.009176, {'classification_error_evaluator': 0.0}
Pass 9, Batch 200, Cost 0.024691, {'classification_error_evaluator': 0.0078125}
Pass 9, Batch 300, Cost 0.025310, {'classification_error_evaluator': 0.015625}
Pass 9, Batch 400, Cost 0.124656, {'classification_error_evaluator': 0.03125}
Test with Pass 9, Cost 0.057880, {'classification_error_evaluator': 0.01759999990463257}

Best pass is 8, testing Avgcost is 0.0480473264802
The classification accuracy is 98.48%

[[0 2 1 7 9 4 3 8 6 5]]
Label of image/infer_7.png is: 0
0:05:38.941172

我原来的图片是我自己PS中画的 图片尺寸是28*28，分辨率是144dpi，图像模式是：RGB color 8bit

并且我尝试了不同的图片，都识别不出来，类似这些，就是猜的貌似也比它强，我怎么感觉我写出来的是人工智障呢？(懵逼.gif)，期待解答。 数字0 数字3 数字6 数字9