29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133 | @dataclass
class DetectionBaseModel(BaseModel):
ontology: DetectionOntology
@abstractmethod
def predict(self, input: str | np.ndarray | Image.Image) -> sv.Detections:
pass
def sahi_predict(self, input: str | np.ndarray | Image.Image) -> sv.Detections:
slicer = sv.InferenceSlicer(callback=self.predict)
return slicer(load_image(input, return_format="cv2"))
def _record_confidence_in_files(
self,
annotations_directory_path: str,
image_names: List[str],
annotations: Dict[str, sv.Detections],
) -> None:
Path(annotations_directory_path).mkdir(parents=True, exist_ok=True)
for image_name in image_names:
detections = annotations[image_name]
yolo_annotations_name, _ = os.path.splitext(image_name)
confidence_path = os.path.join(
annotations_directory_path,
"confidence-" + yolo_annotations_name + ".txt",
)
if detections.confidence is None:
raise ValueError("Expected detections to have confidence values.")
confidence_list = [str(x) for x in detections.confidence.tolist()]
save_text_file(lines=confidence_list, file_path=confidence_path)
print("Saved confidence file: " + confidence_path)
def label(
self,
input_folder: str,
extension: str = ".jpg",
output_folder: str | None = None,
human_in_the_loop: bool = False,
roboflow_project: str | None = None,
roboflow_tags: list[str] = ["autodistill"],
sahi: bool = False,
record_confidence: bool = False,
nms_settings: NmsSetting = NmsSetting.NONE,
) -> sv.DetectionDataset:
"""
Label a dataset with the model.
"""
if output_folder is None:
output_folder = input_folder + "_labeled"
os.makedirs(output_folder, exist_ok=True)
image_paths = glob.glob(input_folder + "/*" + extension)
detections_map = {}
if sahi:
slicer = sv.InferenceSlicer(callback=self.predict)
progress_bar = tqdm(image_paths, desc="Labeling images")
for f_path in progress_bar:
progress_bar.set_description(desc=f"Labeling {f_path}", refresh=True)
image = cv2.imread(f_path)
if sahi:
detections = slicer(image)
else:
detections = self.predict(image)
if nms_settings == NmsSetting.CLASS_SPECIFIC:
detections = detections.with_nms()
if nms_settings == NmsSetting.CLASS_AGNOSTIC:
detections = detections.with_nms(class_agnostic=True)
detections_map[f_path] = detections
dataset = sv.DetectionDataset(
self.ontology.classes(), image_paths, detections_map
)
dataset.as_yolo(
output_folder + "/images",
output_folder + "/annotations",
min_image_area_percentage=0.01,
data_yaml_path=output_folder + "/data.yaml",
)
if record_confidence:
image_names = [os.path.basename(f_path) for f_path in image_paths]
self._record_confidence_in_files(
output_folder + "/annotations", image_names, detections_map
)
split_data(output_folder, record_confidence=record_confidence)
if human_in_the_loop:
roboflow.login()
rf = roboflow.Roboflow()
workspace = rf.workspace()
workspace.upload_dataset(output_folder, project_name=roboflow_project)
print("Labeled dataset created - ready for distillation.")
return dataset
|