Unable to read images simultaneously [in parallels] using pyspark

user_4088201 0
Viewed 334

I have 10 jpeg images in a directory. I want to read all them simultaneously using pyspark. I tried as follows.

from PIL import Image


from pyspark import SparkContext, SparkConf    

conf = SparkConf()
spark = SparkContext(conf=conf)       

files = glob.glob("E:\\tests\\*.jpg")

files_ = spark.parallelize(files)    

arrs = []

for fi in files_.toLocalIterator():      

    im = Image.open(fi)
    data = np.asarray(im)
    arrs.append(data)

img = np.array(arrs)    
print (img.shape)

The code ended without error and printed out img.shape; however, it did not run in parallel. Could you help me?

pyspark apache-spark parallel-processing python-imaging-library
2 Answers

You can use rdd.map to load and transform the pictures in parallel and then collect the rdd into a Python list:

files = glob.glob("E:\\tests\\*.jpg")

file_rdd = spark.parallelize(files)

def image_to_array(path):
    im = Image.open(path)
    data = np.asarray(im)
    return data

array_rdd = file_rdd.map(lambda f: image_to_array(f))
result_list = array_rdd.collect()

result_list is now a list with 10 elements, each element is a numpy.ndarray.

The function image_to_array will be executed on different Spark executors in parallel. If you have a multi-node Spark cluster, you have to make sure that all nodes can access E:\\tests\\.

After collecting the arrays, processing can continue with

img = np.array(result_list, dtype=object)
user_21298010

My solution follows the same idea from werner, but I did only using spark libs:

from pyspark.ml.image import ImageSchema
import numpy as np


df = (spark
      .read
      .format("image")
      .option("pathGlobFilter", "*.jpg")
      .load("your_data_path"))

df = df.select('image.*')

# Pre-caching the required schema. If you remove this line an error will be raised.
ImageSchema.imageFields

# Transforming images to np.array
arrays = df.rdd.map(ImageSchema.toNDArray).collect()

img = np.array(arrays)
print(img.shape)
user_60802760
Related