Image Generator: Output array is read-only

[dalbohn]

Apologies for the cross post from the RStudio Keras github repo, but I thought someone here might have an answer, or might be experiencing an issue similar to mine.

I'm getting an error trying to use a keras image generator.

library(keras)
x <- "test_image.png"
img <- image_load(x)

x <- image_to_array(img)

dim(x) <- c(1,dim(x))

gen_images <- image_data_generator(rescale = 1/255,
                                   preprocessing_function = imagenet_preprocess_input
)

images_iter <- flow_images_from_data (
  x=x, y=NULL,
  generator=gen_images,
  batch_size=1,
  save_to_dir=images_dir,
  save_prefix="aug",
  save_format="jpeg"
)

reticulate::iter_next(images_iter)

Results in the following error: Error in py_iter_next(it, completed) : ValueError: output array is read-only .

The same error occurs when I try to run imagenet_preprocess_input with a directory as input , e.g.,

train_datagen = image_data_generator(
  rescale = 1/255,
  horizontal_flip = TRUE,
  fill_mode = "nearest",
  preprocessing_function = imagenet_preprocess_input
)

train_generator <- flow_images_from_directory(
  "./train",                 
  train_datagen,            
  target_size = c(FLAGS$image_width,FLAGS$image_height),
  batch_size = batch_size,
  class_mode = "categorical",
  follow_links = TRUE
)

Session info:

sessionInfo()
R version 3.6.0 (2019-04-26)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Linux Mint 18.3

Matrix products: default
BLAS:   /usr/lib/libblas/libblas.so.3.6.0
LAPACK: /home/dan/anaconda/envs/r-tensorflow/lib/libmkl_rt.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C               LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8    LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C             LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] keras_2.2.4

loaded via a namespace (and not attached):
 [1] compiler_3.6.0         Matrix_1.2-17          magrittr_1.5           R6_2.3.0               generics_0.0.2        
 [6] tools_3.6.0            whisker_0.3-2          base64enc_0.1-3        Rcpp_1.0.1             reticulate_1.10.0.9003
[11] tensorflow_1.10        grid_3.6.0             zeallot_0.1.0          jsonlite_1.6           tfruns_1.4            
[16] lattice_0.20-38