darkcurrent

Guide:

1. Install Armbian_5.24.161216_Orangepizero_Ubuntu_xenial_3.4.113.img onto a uSD card using Win32DiskImager or Ubuntu Disk Image Writer

2. (Optional) Mount the uSD in Ubuntu Laptop and expand the partition using GParted.

3. Delete everything from uSD except /boot, /lib/modules and /lib/firmware.

4. Mount openwrt-15.05.1-sunxi-root.ext4 on Ubuntu using loopback interface on /mnt/openwrt

darkcurrent

import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
import os
from pathlib import Path

import keras
from keras.preprocessing.image import ImageDataGenerator
from keras.applications import ResNet50
from keras.applications.resnet50 import preprocess_input

darkcurrent

conv_base = ResNet50(include_top=False,
                     weights='imagenet',
                     input_shape=(224, 224, 3))
for layer in conv_base.layers:
    layer.trainable = False
x = conv_base.output
x = layers.GlobalAveragePooling2D()(x)
x = layers.Dense(512, activation='relu')(x)
x = layers.Dense(256, activation='softmax')(x)
predictions = layers.Dense(2, activation='softmax')(x)

darkcurrent

history = model.fit_generator(
    generator=train_generator,
    epochs=15,
    validation_data=validation_generator,
    steps_per_epoch=40,
    validation_steps=20)

plt.plot(history.history['acc'])
plt.plot(history.history['val_acc'])
plt.title('Model doğruluğu')

darkcurrent

Epoch 1/15  40/40 - 43s 1s/step - loss: 0.6341 - acc: 0.8280 - val_loss: 0.6007 - val_acc: 0.9412
Epoch 2/15  40/40 - 29s 716ms/step - loss: 0.5988 - acc: 0.9592 - val_loss: 0.5942 - val_acc: 0.9542
Epoch 3/15  40/40 - 30s 741ms/step - loss: 0.5888 - acc: 0.9830 - val_loss: 0.5916 - val_acc: 0.9412
Epoch 4/15  40/40 - 30s 744ms/step - loss: 0.5843 - acc: 0.9800 - val_loss: 0.5878 - val_acc: 0.9542
Epoch 5/15  40/40 - 30s 741ms/step - loss: 0.5793 - acc: 0.9902 - val_loss: 0.5842 - val_acc: 0.9477
Epoch 6/15  40/40 - 30s 742ms/step - loss: 0.5745 - acc: 0.9913 - val_loss: 0.5821 - val_acc: 0.9477
Epoch 7/15  40/40 - 30s 742ms/step - loss: 0.5691 - acc: 0.9961 - val_loss: 0.5786 - val_acc: 0.9477
Epoch 8/15  40/40 - 30s 743ms/step - loss: 0.5673 - acc: 0.9889 - val_loss: 0.5750 - val_acc: 0.9542
Epoch 9/15  40/40 - 30s 739ms/step - loss: 0.5626 - acc: 0.9937 - val_loss: 0.5712 - val_acc: 0.9608
Epoch 10/15 40/40 - 30s 742ms/step - loss: 0.5581 - acc: 0.9961 - val_loss: 0.5685 - val_acc: 0.9477