1#!/usr/bin/env python 

2# coding=utf-8 

3 

4"""Shenzhen No.3 People’s Hospital dataset for Lung Segmentation 

5 

6The database includes 336 cases with manifestation of tuberculosis, and 326 

7normal cases. It contains a total of 662 images. Image size varies for each 

8X-ray. It is approximately 3K x 3K. One set of ground-truth lung annotations is 

9available for 566 of the 662 images. 

10 

11* Reference: [SHENZHEN-2014]_ 

12* Original resolution (height x width): Approximately 3K x 3K (varies) 

13* Configuration resolution: 512 x 512 (after rescaling) 

14* Split reference: [GAAL-2020]_ 

15* Protocol ``default``: 

16 

17 * Training samples: 396 (including labels) 

18 * Validation samples: 56 (including labels) 

19 * Test samples: 114 (including labels) 

20 

21""" 

22 

23import os 

24 

25import pkg_resources 

26 

27import bob.extension 

28 

29from ..dataset import JSONDataset 

30from ..loader import load_pil_1, load_pil_rgb, make_delayed 

31 

32_protocols = [ 

33 pkg_resources.resource_filename(__name__, "default.json"), 

34] 

35 

36_root_path = bob.extension.rc.get( 

37 "bob.ip.binseg.shenzhen.datadir", os.path.realpath(os.curdir) 

38) 

39 

40 

41def _raw_data_loader(sample): 

42 return dict( 

43 data=load_pil_rgb(os.path.join(_root_path, sample["data"])), 

44 label=load_pil_1(os.path.join(_root_path, sample["label"])), 

45 ) 

46 

47 

48def _loader(context, sample): 

49 # "context" is ignored in this case - database is homogeneous 

50 # we returned delayed samples to avoid loading all images at once 

51 return make_delayed(sample, _raw_data_loader) 

52 

53 

54dataset = JSONDataset( 

55 protocols=_protocols, fieldnames=("data", "label"), loader=_loader 

56) 

57 

58"""Shenzhen CXR dataset object"""