library(png)
library(colorspace)
library(abind)
library(reticulate)
library(keras)

setwd("~/E-books/Coding/Machine Learning with R/Neural Networks/four-shapes")

# Here I create a list of links to retrieve each image
files <- list.files(pattern = ".png$", recursive = TRUE)
filePathFunction <- function(x){
a <- matrix(nrow=length(x))
for(i in 1:length(x)){
conc <- c("~/E-books/Coding/Machine Learning with R/Neural Networks/four-shapes",x[i])
a[i] <- paste(conc, collapse="/")
}
return(a)
}
filePaths <- filePathFunction(files)
image(readPNG(filePaths[1]), useRaster=TRUE, axes=FALSE)


#circle is 0
#square is 1
#star is 2
#triangle is 3

circle <- rep(0,3720)
circle_array <- array(circle, dim=c(3720,1))
square <- rep(1,3765)
square_array <- array(square, dim=c(3765,1))
star <- rep(2,3765)
star_array <- array(star, dim=c(3765,1))
triangle <- rep(3,3720)
triangle_array <- array(triangle, dim=c(3720,1))
labels <- rbind(circle_array,square_array,star_array, triangle_array)



set.seed(101)
ind <- sample(nrow(filePaths),round(0.75*nrow(filePaths)),replace=F)
train_data <- filePaths[ind,]
test_data <- filePaths[-ind,]

train_labels <- labels[ind,]
test_labels <- labels[-ind,]

train_labels <- to_categorical(train_labels)
test_labels <- to_categorical(test_labels)

#trainPathArray and testPathArray reshape each pixel matrix into a 11228x200*200 array and
#3742x200*200 array matrix

trainPathArray <- matrix(ncol=200*200, nrow= length(train_data))
for(i in 1:length(train_data)){
trainPathArray[i,] <- array_reshape(readPNG(train_data[i]), dim= c(1,200*200))
}
testPathArray <- matrix(ncol=200*200, nrow= length(test_data))
for(i in 1:length(test_data)){
testPathArray[i,] <- array_reshape(readPNG(test_data[i]), dim= c(1,200*200))
}


network <- keras_model_sequential() %>%
layer_dense(units = 1000, activation = "relu", input_shape = c(200*200)) %>%
layer_dense(units = 1000, activation = "relu") %>%
layer_dense(units = 4, activation = "softmax")

network %>% compile(
optimizer = "rmsprop",
loss = "categorical_crossentropy",
metrics = c("accuracy")
)
network %>% fit(trainPathArray, train_labels, epochs = 5, batch_size = 12smilie
metrics <- network %>% evaluate(testPathArray, test_labels)