filename = group.filename.encode('utf8')

[julio8a]

Getting this error when generating the tfrecords.

  File "generate_tfrecord.py", line 31
    filename = group.filename.encode('utf8')
                                           ^
IndentationError: unindent does not match any outer indentation level

generating the csv and pbtxt worked fine.

[layshowl]

我也是出现了这个问题，求大神指点

[julio8a]

@layshowl,

I ended up using a different generate_tf file from the repository this fork came from and it worked for me: https://github.com/datitran/raccoon_dataset/blob/master/generate_tfrecord.py

if you have multiple classes you're trying to detect, i.e. Apple, Banana, and Peach then you need to modify line 32 to look like this:

    if row_label == 'Apple':
        return 1
    elif: row_label == 'Banana':
        return 2
    elif: row_label == 'Peach':
        return 3
    else:
        None

[douglasrizzo]

I fixed the indentation problem on master. It came from a PR.