8.1.1.9.1.1.1.1. blueoil.utils.predict_output.output

8.1.1.9.1.1.1.1.1. Module Contents

8.1.1.9.1.1.1.1.1.1. Classes

JsonOutput	Create callable instance to output predictions json object from post processed tensor(np.ndarray).
ImageFromJson	Create callable instance to return list of tuple (file_name, PIL image object) from prediction json.

class blueoil.utils.predict_output.output.JsonOutput(task, classes, image_size, data_format, bench=None)

Create callable instance to output predictions json object from post processed tensor(np.ndarray).

The output predictions json format depends on task type. Please see [Output Data Specification](https://github.com/LeapMind/lmnet/wiki/Output-Data-Specification).

_classification(self, outputs, raw_images, image_files)

_object_detection(self, outputs, raw_images, image_files)

_semantic_segmentation(self, outputs, raw_images, image_files)

_keypoint_detection(self, outputs, raw_images, image_files)

__call__(self, outputs, raw_images, image_files)

Output predictions json object from post processed tensor(np.ndarray).

Parameters

• outputs (np.ndarray or list) – Post processed tensor.

• raw_images (list) – List of np.ndarray of raw (non pre-processed) images.

• image_files (list) – List of image file paths.

class blueoil.utils.predict_output.output.ImageFromJson(task, classes, image_size)

Create callable instance to return list of tuple (file_name, PIL image object) from prediction json.

_classification(self, result_json, raw_images, image_files)

_semantic_segmentation(self, result_json, raw_images, image_files)

_object_detection(self, result_json, raw_images, image_files)

_keypoint_detection(self, result_json, raw_images, image_files)

__call__(self, json_results, raw_images, image_files)