深度学习水果识别

https://gitee.com/dancheng-senior/postgraduate

开发简介

传统的水果图像识别系统的一般过程如下图所示主要工作集中在图像预处理和特征提取阶段。

水果反光、

训练集包含四类apple,orange,banana,mixed(多种水果混合)四类237张图片测试集包含每类图片各两张。

图片集如下图所示。

训练集图片预览

fruits_imageprint(train_fruits)5.2

模型网络结构

model.add(Conv2D(32,(3,3),input_shape

array_image.shape))

model.add(Dense(numberOfClass))

output

categorical_crossentropy,optimizer

rmsprop,metrics

0.3,horizontal_flipTrue,zoom_range

0.3)test_datagen

train_datagen.flow_from_directory(train_dir,target_size

array_image.shape[:2],batch_size

rgb,class_mode

test_datagen.flow_from_directory(test_dir,target_size

array_image.shape[:2],batch_size

rgb,class_mode

train_generator:print(data_batch

shape

,data_batch.shape)print(labels_batch

shape

train_generator,steps_per_epoch

1600

batch_size,epochsepochs,validation_data

test_generator,validation_steps

800

model.save(model_fruits.h5)顺便输出训练曲线

#展示损失模型结果

plt.plot(hist.history[loss],label

Train

plt.plot(hist.history[val_loss],label

Validation

linestyledashed,markeredgecolor

purple,

plt.plot(hist.history[accuracy],label

Train

plt.plot(hist.history[val_accuracy],label

Validation

linestyledashed,markeredgecolor

purple,

ImageDataGenerator,img_to_array,

load_img

0.3,horizontal_flipTrue,zoom_range

0.3)model

load_img(./Test/Apricot/3_100.jpg,target_size(100,100))

plt.imshow(img)

train_datagen.flow_from_directory(train_dir,target_size

array_image.shape[:2],batch_size

rgb,class_mode

print(train_generator.class_indices)​

fig

os.path.isdir(os.path.join(path,

x)),

os.listdir(path)))random_directory

np.random.randint(0,

folders[random_directory]file_path

os.path.join(path,

os.path.isfile(os.path.join(file_path,

np.random.randint(0,

file_names[random_file_index]final_path

image_name)return

valueBLACK)cv2.putText(expanded_image,

predicted:

IMG_COLS)files.append(final_path)true_labels.append(true_label)x

1./255x

batch_size10)predictions.append(classes)class_labels

class_labels

class_labels[predictions[i][0]],

image,

true_labels[i])axes.append(fig.add_subplot(rows,

cols,

i1))plt.imshow(cv2.cvtColor(image,

cv2.COLOR_BGR2RGB))plt.grid(False)plt.axis(off)

plt.show()7

https://gitee.com/dancheng-senior/postgraduate