There is a wrong in data_csv index

[sunshinelike]

[WuJunde]

yes. sorry for leading the misunderstanding. the difference between part1 and 3b is that part1 is only for segmentation, and 3b is for segmentation and diagnosis. I switch it from 3b to 1 in version 2, but both will work fine. you can download part1 data and use this csv anyway.

[sunshinelike]

OK, thank you very much, got, if you could update the readme and file name, that would be better for understanding. And could you share the csv maker script?

[WuJunde]

sure

`import os import pandas as pd

path = '../../data/isic/isic2016' train_data_dir = 'ISBI2016_ISIC_Part1_Training_Data' train_gt_dir = 'ISBI2016_ISIC_Part1_Training_GroundTruth' csv_path_train = '../ISBI2016_ISIC_Part3B_Training_GroundTruth.csv'

col_names = ['img','seg'] df = pd.DataFrame()

for (root,dirs,files) in os.walk(os.path.join(path,train_data_dir), topdown=True): for name in files: preind = name.split('.')[0] segname = preind + '_Segmentation.png' segpath = os.path.join( train_gt_dir, segname) imgpath = os.path.join( train_data_dir, name) data = pd.DataFrame([[imgpath,segpath]], columns = col_names) df = df.append(data, ignore_index=True) df.to_csv(csv_path_train)`

[sunshinelike]

Thank you very much. But when I download part1 data, it has no corresponding segmentation image, the index is also wrong. If I use your script

sure

`import os import pandas as pd

path = '../../data/isic/isic2016' train_data_dir = 'ISBI2016_ISIC_Part1_Training_Data' train_gt_dir = 'ISBI2016_ISIC_Part1_Training_GroundTruth' csv_path_train = '../ISBI2016_ISIC_Part3B_Training_GroundTruth.csv'

col_names = ['img','seg'] df = pd.DataFrame()

for (root,dirs,files) in os.walk(os.path.join(path,train_data_dir), topdown=True): for name in files: preind = name.split('.')[0] segname = preind + '_Segmentation.png' segpath = os.path.join( train_gt_dir, segname) imgpath = os.path.join( train_data_dir, name) data = pd.DataFrame([[imgpath,segpath]], columns = col_names) df = df.append(data, ignore_index=True) df.to_csv(csv_path_train)`

[WuJunde]

download training ground truth, which is the segmentation

[sunshinelike]

download training ground truth, which is the segmentation Oh, I see. Thank you very much, best wishes.