prediction-filtering-optimization/phase_error_visualization.py at main · Andrii-Rudavskyi/prediction-filtering-optimization · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
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
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
import os
import seaborn as sn
from helper2 import calculate_delay_error, calculate_predicted_error, predict, extract_noise_2, calculate_phase_error, PredictionModel, PredictionModelType

dataFolder = './data/'
imagesFolder = './images/'

resolutionWidth = 3840
resolutionHeight = 2160
pixelPitch = 0.08964

screenWidth = resolutionWidth * pixelPitch
screenHeight = resolutionHeight * pixelPitch

#Camera position with respect to the screen TODO: load it from the product config data
xc = screenWidth / 2
yc = 0

latency = 0.1

# Focal distance to convert to angular space. TODO: Load it properly from intrinsics (product/device specific).
fx = 330
fy = 330

#selection of the prediction model
predictionModel = PredictionModel(PredictionModelType.PolynomialFit)
predictionModel.setBufferSize(5)
predictionModel.setPolynomialOrder(2)

for filename in os.listdir(dataFolder):
    if filename.endswith('cnsdk_blink_raw_data_trace.csv'):
        eyetrackerdata = pd.read_csv(dataFolder + filename)
        imudata = pd.read_csv(dataFolder + filename)

        eyetrackerdata_timestapms = eyetrackerdata[' cameraTimestamp']
        #Start trace from zero
        eyetrackerdata_timestapms = 0.001 * (eyetrackerdata_timestapms - eyetrackerdata_timestapms[0])

        #Convert to angular space
        leftEyeX = eyetrackerdata[' leftEyeX'] / fx
        leftEyeY = eyetrackerdata[' leftEyeY'] / fy
        leftEye3DZ = eyetrackerdata[' leftEye3DZ']

        signal = [leftEyeX, leftEyeY, leftEye3DZ]
        signal_label = ['leftEyeX', 'leftEyeY', 'leftEye3DZ']

        filtered_time = []
        filtered_signal = []
        noise = []
        delay_error = []

        predicted_error = []
        predicted = []
        ground_truth = []

        #Calculate noise, filtered signal and delay error for x2d,y2d and z
        for i in range(0, 3):
            filtered_time_tmp, filtered_signal_tmp, noise_tmp = extract_noise_2(signal[i], eyetrackerdata_timestapms)
            delay_error_tmp = calculate_delay_error(filtered_time_tmp, filtered_signal_tmp, predictionTime=latency)
            filtered_time.append(filtered_time_tmp)
            filtered_signal.append(filtered_signal_tmp)
            noise.append(noise_tmp)
            delay_error.append(delay_error_tmp)
            predicted_tmp = predict(eyetrackerdata_timestapms, signal[i], predictionModel, predcictionTime=latency)
            predicted.append(predicted_tmp)

            ground_truth_tmp, predicted_error_tmp = calculate_predicted_error(eyetrackerdata_timestapms, signal[i], predicted_tmp,
                                                                      predictionTime=latency)
            predicted_error.append(predicted_error_tmp)
            ground_truth.append(ground_truth_tmp)

        n = 100

        size = int(resolutionWidth/n), int(resolutionHeight/n)
        error_heatmap = np.zeros(size)
        predicted_phase_error_heatmap = np.zeros(size)

        for k in range(0, len(delay_error[0])):
            for i in range(0, int(resolutionWidth / n)):
                for j in range(0, int(resolutionHeight / n)):
                    phase_error = calculate_phase_error(xc, yc, delay_error[2][k], delay_error[0][k], delay_error[1][k], filtered_signal[2][k], i * n * pixelPitch, j * n * pixelPitch)
                    error_heatmap[i, j] = phase_error

                    phase_error = calculate_phase_error(xc, yc, predicted_error[2][k], predicted_error[0][k], predicted_error[1][k], filtered_signal[2][k], i * n * pixelPitch, j * n * pixelPitch)
                    predicted_phase_error_heatmap[i, j] = phase_error

            frameIDstr = 0
            if k < 10:
                frameIDstr = '000' + str(k)
            if (k >= 10 and k < 100):
                frameIDstr = '00' + str(k)
            if (k >= 100 and k < 1000):
                frameIDstr = '0' + str(k)
            if (k >= 1000):
                frameIDstr = str(k)

            plt.figure(figsize=(10, 8))
            for l in range(0, 3):
                plt.subplot(3, 2, l + 1)
                plt.plot(filtered_time[l], filtered_signal[l])
                plt.title(str(signal_label[l]))
                plt.plot(filtered_time[l], delay_error[l])
                plt.plot(eyetrackerdata_timestapms, ground_truth[l])
                plt.plot(eyetrackerdata_timestapms, predicted[l])
                plt.plot(eyetrackerdata_timestapms, predicted_error[l])
                plt.legend(['Signal', 'Delay Error', 'Ground truth', 'Predicted', 'Predicted error'])
                plt.axvline(x=filtered_time[l][k])

            plt.subplot(3, 2, 4)
            sn.heatmap(error_heatmap.transpose(), vmin=-0.5, vmax=0.5)
            plt.subplot(3, 2, 5)
            sn.heatmap(predicted_phase_error_heatmap.transpose(), vmin=-0.5, vmax=0.5)

            plt.savefig(imagesFolder + 'frame' + frameIDstr + '.png')
            #plt.show()

# fps = 60
# output_dir = './images'
# # Generate the MP4 video using ffmpeg
# ffmpeg_command = f"ffmpeg -y -r {fps} -i {output_dir}/frame%04d.png -vcodec libx264 -pix_fmt yuv420p {'./' + 'video.mp4'}"
# exit_status = os.system(ffmpeg_command)