Coverage for src/bob/bio/base/database/atnt.py: 50%

1#!/usr/bin/env python 

2""" 

3ATNT database implementation 

4""" 

5 

6from pathlib import Path 

7 

8from sklearn.pipeline import make_pipeline 

9 

10import bob.io.base 

11 

12from bob.bio.base.database.utils import download_file, md5_hash 

13 

14from . import CSVDatabase, FileSampleLoader 

15 

16 

17class AtntBioDatabase(CSVDatabase): 

18 """ 

19 The AT&T (aka ORL) database of faces 

20 (http://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html). This 

21 class defines a simple protocol for training, enrollment and probe by 

22 splitting the few images of the database in a reasonable manner. Due to the 

23 small size of the database, there is only a 'dev' group, and I did not 

24 define an 'eval' group. 

25 """ 

26 

27 category = "base" 

28 dataset_protocols_urls = [ 

29 "https://www.idiap.ch/software/bob/databases/latest/base/atnt-f529acef.tar.gz", 

30 "http://www.idiap.ch/software/bob/databases/latest/base/atnt-f529acef.tar.gz", 

31 ] 

32 dataset_protocols_checksum = "f529acef" 

33 

34 def __init__( 

35 self, 

36 protocol="idiap_protocol", 

37 dataset_original_directory=None, 

38 **kwargs, 

39 ): 

40 """Custom init to download the raw data files.""" 

41 

42 # Download the raw data (or use a cached local version) 

43 if dataset_original_directory is None: 

44 path = download_file( 

45 urls=[ 

46 "http://www.idiap.ch/software/bob/data/bob/att_faces.zip" 

47 ], 

48 destination_sub_directory="datasets/atnt", 

49 destination_filename="atnt_faces.zip", 

50 checksum="6efb25cb0d40755e9492b9c012e3348d", 

51 checksum_fct=md5_hash, 

52 extract=True, 

53 ) 

54 dataset_original_directory = Path(path).as_posix() 

55 

56 super().__init__( 

57 name="atnt", 

58 protocol=protocol, 

59 transformer=make_pipeline( 

60 FileSampleLoader( 

61 data_loader=bob.io.base.load, 

62 dataset_original_directory=dataset_original_directory, 

63 extension=".pgm", 

64 ), 

65 ), 

66 **kwargs, 

67 ) 

68 # just expose original_directory for backward compatibility of tests 

69 self.original_directory = dataset_original_directory 

70 self.original_extension = ".pgm" 

71 

72 # define an objects method for compatibility with the old tests 

73 def objects( 

74 self, model_ids=None, groups=None, purposes=None, protocol=None 

75 ): 

76 samples = [] 

77 

78 if groups is None: 

79 groups = self.groups() 

80 

81 if "train" in groups or "world" in groups: 

82 samples += self.background_model_samples() 

83 

84 if purposes is None: 

85 purposes = ("enroll", "probe") 

86 

87 if "enroll" in purposes: 

88 samples += self.references() 

89 

90 if "probe" in purposes: 

91 samples += self.probes() 

92 

93 if model_ids: 

94 samples = [s for s in samples if s.template_id in model_ids] 

95 

96 # create the old attributes 

97 for s in samples: 

98 s.client_id = s.template_id 

99 s.path = s.id = s.key 

100 

101 return samples