Problems With Reading In Csv Values Into 2d Numpy Array

I am having issues reading in values from a saved csv file. This is part of the csv file that I have: 000000216739.jpg, 224, [ 0. 0. 0. 0. 36. 44. 4. 0. 0. 0. 0. 0. 0.

Solution 1:

I presume your data is like this; filename,nrows,[imrow1],[imrow2],[imrow3], ...,[imrow_nrows] then do the following to extract the image from your data. Let us know if your data is formatted differently.

with open(CSVFilepath) as f:
  reader = csv.reader(f,delimiter=',')

  for row in reader:
    fname=row[0]
    nrows=int(row[1])
    imdata=np.array(row[2:]) #Get from the 2 element to the end. Could also do row[:-nrows]
  #might need to make each element an int or float.        
  ###reshape if that's what you need and do something with the image

Solution 2:

Your file is not a correct csv file, and you should not read it like a csv file.

The linebreak in csv file represent a new row whereas it's apparent that in your file they don't mean that - you want to read the numbers inside [ and ] but they're not delimited properly.

A way to dissect this file would be to

with open(file,'r') as fin:
    f = fin.readlines()
    f = ' '.join(f) # remove newlines
    listrows = f.split('[')
    listrows = [l.split(']')[0] for l in listrows] # Get string between '[' and ']'
    matrix = [row.split('.') for row in listrows] # This is now a 2D matrix
    final = [[int(e.replace(' ','')) for e in row] for row in matrix] # Here goes your final matrix

I used list comprehension extensively so this does not go to 30 lines. Try running this.