FORMAT_CONVERSION_MAXCOMPUTE/idc/model.py

# -*- coding: utf-8 -*-
"""
Created on Tue Jun 21 10:53:51 2022
model
@author: fangjiasheng
"""
import os
import sys
sys.path.append(os.path.dirname(os.path.abspath(__file__)))

from keras.layers import Lambda, Dense, Reshape, Conv2D, BatchNormalization, LeakyReLU, Masking, MaxPool2D, \
    MaxPooling2D, UpSampling2D, concatenate, Activation, GlobalAveragePooling2D, DepthwiseConv2D, Add
from keras import layers, models, Sequential, Input, Model
import keras.backend as K


def direction_model(input_shape, output_shape):
    model = cnn_model(input_shape, output_shape)
    # print(input_shape, output_shape)
    # model = mobile_net_v3_tiny(input_shape, output_shape)
    # model = fpn(input_shape, output_shape)
    # model.summary(line_length=100)
    return model


def cnn_model(input_shape, output_shape):
    conv_num = 6

    # Input
    _input = Input(shape=input_shape, dtype="float32", name="input")

    conv = Conv2D(16, (3, 3), padding='same')(_input)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    max_pool = MaxPool2D()(relu)
    for i in range(conv_num):
        conv = Conv2D(16, (3, 3), padding='same')(max_pool)
        bn = BatchNormalization()(conv)
        relu = LeakyReLU(alpha=0.)(bn)
        # conv = Conv2D(32, (1, 1), padding='same')(relu)
        # bn = BatchNormalization()(conv)
        # relu = LeakyReLU(alpha=0.)(bn)
        max_pool = MaxPool2D()(relu)
    # conv = Conv2D(16, (3, 3), padding='same')(max_pool)
    # bn = BatchNormalization()(conv)
    # relu = LeakyReLU(alpha=0.)(bn)
    max_pool = MaxPool2D((6, 6))(relu)

    dense = layers.Dense(output_shape, activation='softmax')(max_pool)
    squeeze = Lambda(lambda x: K.squeeze(x, axis=1))(dense)
    squeeze = Lambda(lambda x: K.squeeze(x, axis=1))(squeeze)

    model = Model(inputs=_input, outputs=squeeze)
    return model


def cnn_model_240314(input_shape, output_shape):
    conv_num = 5

    # Input
    _input = Input(shape=input_shape, dtype="float32", name="input")

    conv = Conv2D(16, (3, 3), padding='same')(_input)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    max_pool = MaxPool2D()(relu)
    for i in range(conv_num):
        conv = Conv2D(16, (3, 3), padding='same')(max_pool)
        bn = BatchNormalization()(conv)
        relu = LeakyReLU(alpha=0.)(bn)
        # conv = Conv2D(32, (1, 1), padding='same')(relu)
        # bn = BatchNormalization()(conv)
        # relu = LeakyReLU(alpha=0.)(bn)
        max_pool = MaxPool2D()(relu)
    # conv = Conv2D(16, (3, 3), padding='same')(max_pool)
    # bn = BatchNormalization()(conv)
    # relu = LeakyReLU(alpha=0.)(bn)
    max_pool = MaxPool2D((6, 6))(relu)

    dense = layers.Dense(output_shape, activation='softmax')(max_pool)
    squeeze = Lambda(lambda x: K.squeeze(x, axis=1))(dense)
    squeeze = Lambda(lambda x: K.squeeze(x, axis=1))(squeeze)

    model = Model(inputs=_input, outputs=squeeze)
    return model


def fpn(input_shape, output_shape):
    # Input
    _input = Input(shape=input_shape, dtype="float32", name="input")

    # 192 -> 96
    conv = Conv2D(8, (3, 3), padding='same')(_input)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    conv = Conv2D(8, (3, 3), padding='same')(relu)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    max_pool_1 = MaxPool2D()(relu)

    # 96 -> 48
    conv = Conv2D(16, (3, 3), padding='same')(max_pool_1)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    conv = Conv2D(16, (3, 3), padding='same')(relu)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    max_pool_2 = MaxPool2D()(relu)

    # 48 -> 24
    conv = Conv2D(32, (3, 3), padding='same')(max_pool_2)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    conv = Conv2D(32, (3, 3), padding='same')(relu)
    bn = BatchNormalization()(conv)
    relu = LeakyReLU(alpha=0.)(bn)
    max_pool_3 = MaxPool2D()(relu)

    #
    conv_pre = Conv2D(32, (1, 1))(max_pool_2)
    up_sample = UpSampling2D((2, 2))(max_pool_3)
    add_1 = Add()([conv_pre, up_sample])

    conv_pre = Conv2D(16, (1, 1))(max_pool_1)
    up_sample = UpSampling2D((2, 2))(max_pool_2)
    add_2 = Add()([conv_pre, up_sample])

    global_pool_1 = GlobalAveragePooling2D()(max_pool_3)
    global_pool_2 = GlobalAveragePooling2D()(add_1)
    global_pool_3 = GlobalAveragePooling2D()(add_2)

    reshape_1 = Reshape((1, 1, 32))(global_pool_1)
    reshape_2 = Reshape((1, 1, 32))(global_pool_2)
    reshape_3 = Reshape((1, 1, 16))(global_pool_3)

    conv_1 = Conv2D(64, (1, 1), padding='same')(reshape_1)
    conv_2 = Conv2D(64, (1, 1), padding='same')(reshape_2)
    conv_3 = Conv2D(64, (1, 1), padding='same')(reshape_3)

    conv_1 = Conv2D(output_shape, (1, 1), padding='same')(conv_1)
    conv_2 = Conv2D(output_shape, (1, 1), padding='same')(conv_2)
    conv_3 = Conv2D(output_shape, (1, 1), padding='same')(conv_3)

    reshape_1 = Reshape((output_shape,))(conv_1)
    reshape_2 = Reshape((output_shape,))(conv_2)
    reshape_3 = Reshape((output_shape,))(conv_3)

    add = Add()([reshape_1, reshape_2, reshape_3])
    softmax = Activation('softmax')(add)

    model = Model(_input, softmax)
    return model


def tiny_unet(input_shape, output_shape):
    inputs = Input(shape=input_shape)
    use_bias = False

    # 128
    down1 = Conv2D(16, (3, 3), padding='same', use_bias=use_bias)(inputs)
    down1 = BatchNormalization()(down1)
    down1 = LeakyReLU(alpha=0.1)(down1)
    down1 = Conv2D(16, (1, 1), padding='same', use_bias=use_bias)(down1)
    down1 = BatchNormalization()(down1)
    down1 = LeakyReLU(alpha=0.1)(down1)
    down1_pool = MaxPooling2D((2, 2), strides=(2, 2))(down1)

    # 64
    down2 = Conv2D(32, (3, 3), padding='same', use_bias=use_bias)(down1_pool)
    down2 = BatchNormalization()(down2)
    down2 = LeakyReLU(alpha=0.1)(down2)
    down2 = Conv2D(32, (1, 1), padding='same', use_bias=use_bias)(down2)
    down2 = BatchNormalization()(down2)
    down2 = LeakyReLU(alpha=0.1)(down2)
    down2_pool = MaxPooling2D((2, 2), strides=(2, 2))(down2)

    # 32
    down3 = Conv2D(64, (3, 3), padding='same', use_bias=use_bias)(down2_pool)
    down3 = BatchNormalization()(down3)
    down3 = LeakyReLU(alpha=0.1)(down3)
    down3 = Conv2D(64, (1, 1), padding='same', use_bias=use_bias)(down3)
    down3 = BatchNormalization()(down3)
    down3 = LeakyReLU(alpha=0.1)(down3)
    down3_pool = MaxPooling2D((2, 2), strides=(2, 2))(down3)

    # 16
    center = Conv2D(64, (3, 3), padding='same', use_bias=use_bias)(down3_pool)
    center = BatchNormalization()(center)
    center = LeakyReLU(alpha=0.1)(center)
    center = Conv2D(64, (1, 1), padding='same', use_bias=use_bias)(center)
    center = BatchNormalization()(center)
    center = LeakyReLU(alpha=0.1)(center)

    # 32
    up3 = UpSampling2D((2, 2))(center)
    up3 = concatenate([down3, up3], axis=3)
    up3 = Conv2D(64, (3, 3), padding='same', use_bias=use_bias)(up3)
    up3 = BatchNormalization()(up3)
    up3 = LeakyReLU(alpha=0.1)(up3)
    up3 = Conv2D(64, (1, 1), padding='same', use_bias=use_bias)(up3)
    up3 = BatchNormalization()(up3)
    up3 = LeakyReLU(alpha=0.1)(up3)

    # 64
    up2 = UpSampling2D((2, 2))(up3)
    up2 = concatenate([down2, up2], axis=3)
    up2 = Conv2D(32, (3, 3), padding='same', use_bias=use_bias)(up2)
    up2 = BatchNormalization()(up2)
    up2 = LeakyReLU(alpha=0.1)(up2)
    up2 = Conv2D(32, (1, 1), padding='same', use_bias=use_bias)(up2)
    up2 = BatchNormalization()(up2)
    up2 = LeakyReLU(alpha=0.1)(up2)

    # 128
    up1 = UpSampling2D((2, 2))(up2)
    up1 = K.concatenate([down1, up1], axis=3)
    up1 = Conv2D(16, (3, 3), padding='same', use_bias=use_bias)(up1)
    up1 = BatchNormalization()(up1)
    up1 = LeakyReLU(alpha=0.1)(up1)
    up1 = Conv2D(16, (1, 1), padding='same', use_bias=use_bias)(up1)
    up1 = BatchNormalization()(up1)
    up1 = LeakyReLU(alpha=0.1)(up1)

    classify = Conv2D(output_shape, (1, 1), activation='softmax')(up1)
    model = Model(inputs=inputs, outputs=classify)
    return model