BEAT - zfang/livdet

This algorithm is a legacy one. The API has changed since its implementation. New versions and forks will need to be updated.

This algorithm is an analyzer. It can only be used on analysis blocks.

Endpoint Groups 1

Algorithms have at least one input and one output. All algorithm endpoints are organized in groups. Groups are used by the platform to indicate which inputs and outputs are synchronized together. The first group is automatically synchronized with the channel defined by the block in which the algorithm is deployed.

Group: group

Endpoint Name	Data Format	Nature
category	system/integer/1	Input
scores	system/integer/1	Input

Results 25

Analyzers may produce any number of results. Once experiments using this analyzer are done, you may display the results or filter experiments using criteria based on them.

Name	Type
numberNotProcessedT1Samples	int32
numberNotProcessedT5Samples	int32
totalNumberT1Samples	int32
ACER	float32
rocT1T5	plot/isoroc/1
rocT1T3	plot/isoroc/1
numberCorrectT5Samples	int32
rocT1T6	plot/isoroc/1
rocT1T2	plot/isoroc/1
numberCorrectT3Samples	int32
totalNumberT5Samples	int32
numberNotProcessedT6Samples	int32
numberNotProcessedT4Samples	int32
numberCorrectT4Samples	int32
numberCorrectT1Samples	int32
totalNumberT2Samples	int32
numberNotProcessedT2Samples	int32
numberCorrectT2Samples	int32
rocAll	plot/isoroc/1
totalNumberT4Samples	int32
totalNumberT6Samples	int32
numberNotProcessedT3Samples	int32
numberCorrectT6Samples	int32
totalNumberT3Samples	int32
rocT1T4	plot/isoroc/1

import numpy as np
import bob.measure

class Algorithm:
    def __init__(self):
        self.T1 = None
        self.T2 = None
        self.T3 = None
        self.T4 = None
        self.T5 = None
        self.T6 = None
        self.allspoofs = None
        self.correct_T1 = []
        self.correct_T2 = []
        self.correct_T3 = []
        self.correct_T4 = []
        self.correct_T5 = []
        self.correct_T6 = []
        self.incorrect_T1 = []
        self.incorrect_T2 = []
        self.incorrect_T3 = []
        self.incorrect_T4 = []
        self.incorrect_T5 = []
        self.incorrect_T6 = []
        self.notprocessed_T1 = []
        self.notprocessed_T2 = []
        self.notprocessed_T3 = []
        self.notprocessed_T4 = []
        self.notprocessed_T5 = []
        self.notprocessed_T6 = []
        self.acer = None
        self.n_bins = 50
        self.threshold = 50
    def process(self, inputs, outputs):
        # accumulate the test scores
        data_test = inputs['scores'].data.value
        label_test = inputs['category'].data.value
        
        if label_test == 1:
            if data_test < 0 or data_test > 100:
                self.notprocessed_T1.append(data_test)
            elif data_test > self.threshold:
                self.correct_T1.append(data_test)
            else:
                self.incorrect_T1.append(data_test)
        elif label_test == 2:
            if data_test < 0 or data_test > 100:
                self.notprocessed_T2.append(data_test)
            elif data_test <= self.threshold:
                self.correct_T2.append(data_test)
            else:
                self.incorrect_T2.append(data_test)
        elif label_test == 3:
            if data_test < 0 or data_test > 100:
                self.notprocessed_T3.append(data_test)
            elif data_test <= self.threshold:
                self.correct_T3.append(data_test)
            else:
                self.incorrect_T3.append(data_test)
        elif label_test == 4:
            if data_test < 0 or data_test > 100:
                self.notprocessed_T4.append(data_test)
            elif data_test <= self.threshold:
                self.correct_T4.append(data_test)
            else:
                self.incorrect_T4.append(data_test)
        elif label_test == 5:
            if data_test < 0 or data_test > 100:
                self.notprocessed_T5.append(data_test)
            elif data_test <= self.threshold:
                self.correct_T5.append(data_test)
            else:
                self.incorrect_T5.append(data_test)
        elif label_test == 6:
            if data_test < 0 or data_test > 100:
                self.notprocessed_T6.append(data_test)
            elif data_test <= self.threshold:
                self.correct_T6.append(data_test)
            else:
                self.incorrect_T6.append(data_test)

# once all values are received, evaluate the scores
        if not(inputs.hasMoreData()):
            self.T1 = np.array(self.correct_T1+self.incorrect_T1)
            self.T2 = np.array(self.correct_T2+self.incorrect_T2)
            self.T3 = np.array(self.correct_T3+self.incorrect_T3)
            self.T4 = np.array(self.correct_T4+self.incorrect_T4)
            self.T5 = np.array(self.correct_T5+self.incorrect_T5)
            self.T6 = np.array(self.correct_T6+self.incorrect_T6)
            self.allspoofs = np.array(self.correct_T2+self.incorrect_T2+self.correct_T3+self.incorrect_T3+self.correct_T4+self.incorrect_T4+self.correct_T5+self.incorrect_T5+self.correct_T6+self.incorrect_T6)

apcer = np.float32(len(self.incorrect_T2+self.incorrect_T3+self.incorrect_T4+self.incorrect_T5+self.incorrect_T6)/len(self.allspoofs))
            bpcer = np.float32(len(self.incorrect_T1)/len(self.T1))
            self.acer = np.float32((apcer+bpcer)/2)

# test_hist_pos, test_bin_pos = np.histogram(self.lives, self.n_bins)
            # test_hist_neg, test_bin_neg = np.histogram(self.spoofs, self.n_bins)
            
            # eer_threshold = bob.measure.eer_threshold(self.spoofs.astype(np.double), self.lives.astype(np.double))
            # far_test, frr_test = bob.measure.farfrr(self.negatives_test, self.lives, eer_threshold)

rocAll  = bob.measure.roc(self.T1.astype(np.double), self.allspoofs.astype(np.double), 100)
            rocT1T2 = bob.measure.roc(self.T1.astype(np.double), self.T2.astype(np.double), 100)
            rocT1T3 = bob.measure.roc(self.T1.astype(np.double), self.T3.astype(np.double), 100)
            rocT1T4 = bob.measure.roc(self.T1.astype(np.double), self.T4.astype(np.double), 100)
            rocT1T5 = bob.measure.roc(self.T1.astype(np.double), self.T5.astype(np.double), 100)
            rocT1T6 = bob.measure.roc(self.T1.astype(np.double), self.T6.astype(np.double), 100)

# writes the output back to the platform
            outputs.write({
                'totalNumberT1Samples': np.int32(len(self.T1)),
                'totalNumberT2Samples': np.int32(len(self.T2)),
                'totalNumberT3Samples': np.int32(len(self.T3)),
                'totalNumberT4Samples': np.int32(len(self.T4)),
                'totalNumberT5Samples': np.int32(len(self.T5)),
                'totalNumberT6Samples': np.int32(len(self.T6)),
                'numberCorrectT1Samples': np.int32(len(self.correct_T1)),
                'numberCorrectT2Samples': np.int32(len(self.correct_T2)),
                'numberCorrectT3Samples': np.int32(len(self.correct_T3)),
                'numberCorrectT4Samples': np.int32(len(self.correct_T4)),
                'numberCorrectT5Samples': np.int32(len(self.correct_T5)),
                'numberCorrectT6Samples': np.int32(len(self.correct_T6)),
                'numberNotProcessedT1Samples': np.int32(len(self.notprocessed_T1)),
                'numberNotProcessedT2Samples': np.int32(len(self.notprocessed_T2)),
                'numberNotProcessedT3Samples': np.int32(len(self.notprocessed_T3)),
                'numberNotProcessedT4Samples': np.int32(len(self.notprocessed_T4)),
                'numberNotProcessedT5Samples': np.int32(len(self.notprocessed_T5)),
                'numberNotProcessedT6Samples': np.int32(len(self.notprocessed_T6)),
                'ACER': np.float32(self.acer),
                'rocAll': {
                    "data":
                        [
                            {
                                "label": 'rocAll',
                                "false_positives": rocAll[1],
                                "false_negatives": rocAll[0],
                                "number_of_positives": np.uint64(len(self.T1)),
                                "number_of_negatives": np.uint64(len(self.allspoofs))
                            }
                        ]
                }, 
                'rocT1T2': {
                    "data":
                        [
                            {
                                "label": 'rocT1T2',
                                "false_positives": rocT1T2[1],
                                "false_negatives": rocT1T2[0],
                                "number_of_positives": np.uint64(len(self.T1)),
                                "number_of_negatives": np.uint64(len(self.T2))
                            }
                        ]
                }, 
                'rocT1T3': {
                    "data":
                        [
                            {
                                "label": 'rocT1T3',
                                "false_positives": rocT1T3[1],
                                "false_negatives": rocT1T3[0],
                                "number_of_positives": np.uint64(len(self.T1)),
                                "number_of_negatives": np.uint64(len(self.T3))
                            }
                        ]
                }, 
                'rocT1T4': {
                    "data":
                        [
                            {
                                "label": 'rocT1T4',
                                "false_positives": rocT1T4[1],
                                "false_negatives": rocT1T4[0],
                                "number_of_positives": np.uint64(len(self.T1)),
                                "number_of_negatives": np.uint64(len(self.T4))
                            }
                        ]
                }, 
                'rocT1T5': {
                    "data":
                        [
                            {
                                "label": 'rocT1T5',
                                "false_positives": rocT1T5[1],
                                "false_negatives": rocT1T5[0],
                                "number_of_positives": np.uint64(len(self.T1)),
                                "number_of_negatives": np.uint64(len(self.T5))
                            }
                        ]
                }, 
                'rocT1T6': {
                    "data":
                        [
                            {
                                "label": 'rocT1T6',
                                "false_positives": rocT1T6[1],
                                "false_negatives": rocT1T6[0],
                                "number_of_positives": np.uint64(len(self.T1)),
                                "number_of_negatives": np.uint64(len(self.T6))
                            }
                        ]
                }
            })

return True

The code for this algorithm in Python
The ruler at 80 columns indicate suggested POSIX line breaks (for readability).
The editor will automatically enlarge to accomodate the entirety of your input
Use keyboard shortcuts for search/replace and faster editing. For example, use Ctrl-F (PC) or Cmd-F (Mac) to search through this box

Could not find any documentation for this object.

No experiments are using this algorithm.

This algorithm was never executed.

algorithms zfang livdet_analyzer 1

Endpoint Groups 1

Group: group

Results 25

algorithms

zfang

livdet_analyzer

1