Appendix A. Specifications#
Classification performance#
Classification performance was measured on a two datasets:
- Cooperative dataset (containing 20K images from various sources obtained at several banks);
- Non cooperative dataset (containing 20K).
The two tables below contain true positive rates corresponding to select false positive rates.
"Classification performance @ low FPR on cooperative dataset"
FPR | TPR CNN 58 | TPR CNN 59 | TPR CNN 59m | TPR CNN 60 | TPR CNN 60m | TPR CNN 62 | TPR CNN 65 |
---|---|---|---|---|---|---|---|
10^-7^ | 0.9910 | 0.9911 | 0.9809 | 0.9917 | 0.979 | 0.9916 | 0.9909 |
10^-6^ | 0.9916 | 0.9915 | 0.9876 | 0.9917 | 0.989 | 0.9917 | 0.9950 |
10^-5^ | 0.9918 | 0.9919 | 0.9904 | 0.9919 | 0.990 | 0.9918 | 0.9976 |
10^-4^ | 0.9919 | 0.9921 | 0.9915 | 0.9921 | 0.991 | 0.9920 | 0.9988 |
"Classification performance @ low FPR on non cooperative dataset"
FPR | TPR CNN 58 | TPR CNN 59 | TPR CNN 59m | TPR CNN 60 | TPR CNN 60m | TPR CNN 62 | TPR CNN 65 |
---|---|---|---|---|---|---|---|
10^-7^ | 0.9834 | 0.9850 | 0.9059 | 0.9862 | 0.9279 | 0.9909 | 0.9909 |
10^-6^ | 0.9914 | 0.9907 | 0.9454 | 0.9931 | 0.9523 | 0.9950 | 0.9950 |
10^-5^ | 0.9954 | 0.9956 | 0.9705 | 0.9967 | 0.9752 | 0.9976 | 0.9976 |
10^-4^ | 0.9983 | 0.9983 | 0.9868 | 0.9987 | 0.9888 | 0.9988 | 0.9988 |
Runtime performance for CentOS Linux environment#
Face detection performance depends on input image parameters such as resolution and bit depth as well as the size of the detected face.
Input data characteristics:
- Image resolution: 1920x1080px;
- Image format: 24 BPP RGB;
Performance measurements are presented for CPU, GPU and NPU execution modes in tables below. Measured values are averages of at least 100 experiments.
Estimated values of memory consumption are also presented for CPU and GPU. These values are highly depend on the input data and the conditions of the experiment.
The results for minimum batch size and optimal batch size are shown in the tables below. All the intermediate and non-optimal values are omitted.
Face detections are performed using FaceDetV3 NN.
All types of face detection and redetect performed with capturing bounding boxes and 5 facial landmarks.
CPU performance#
Benchmarking for CPU was performed on the server with the following hardware configuration:
CPU:
- Intel(R) Xeon(R) Silver 4210 CPU @ 2.20GHz;
- CPU(s): 40
- Thread(s) per core: 2
- Core(s) per socket: 10
- Socket(s): 2
- NUMA node(s): 2
- CPU with AVX2 instruction set was used
OS: CentOS Linux release 8.3.2011
RAM: 128 GB DDR4 (Clock Speed: 2133 MHz)
In experiments listed in tables below face detection and descriptor extraction algorithms used all available CPU cores, whereas matching performance is specified per-core.
Descriptor matching is only implemented on CPU.
CPU. Detector performance#
The table below shows the performance of FaceDetV3 Detector on the CPU.
Measurement | CPU threads | BatchSize | Average (ms) | RAM Memory (Mb) |
---|---|---|---|---|
Detector (minFaceSize=20) | 1 | 1 | 373.92 | 1889.0 |
Detector (minFaceSize=20) | 8 | 1 | 152.73 | 2076.0 |
Detector (minFaceSize=20) | 8 | 4 | 147.26 | 4411.0 |
Detector (minFaceSize=20) | 8 | 8 | 148.32 | 7329.0 |
Detector (minFaceSize=50) | 1 | 1 | 63.23 | 1261.0 |
Detector (minFaceSize=50) | 8 | 1 | 27.52 | 1482.0 |
Detector (minFaceSize=50) | 8 | 4 | 23.43 | 1810.0 |
Detector (minFaceSize=50) | 8 | 8 | 24.61 | 2358.0 |
Detector (minFaceSize=90) | 1 | 1 | 23.11 | 1184.0 |
Detector (minFaceSize=90) | 8 | 1 | 11.62 | 1364.0 |
Detector (minFaceSize=90) | 8 | 4 | 8.03 | 1470.0 |
Detector (minFaceSize=90) | 8 | 8 | 8.23 | 1748.0 |
Redetect | 1 | 1 | 0.63 | 1252.0 |
Redetect | 8 | 1 | 0.83 | 1284.0 |
Redetect | 8 | 4 | 0.32 | 1673.0 |
Redetect | 8 | 8 | 0.25 | 2153.0 |
FaceLandmarks5Detector | 1 | 1 | 0.22 | 1225.0 |
FaceLandmarks5Detector | 8 | 1 | 0.37 | 1225.0 |
FaceLandmarks5Detector | 8 | 8 | 0.09 | 1226.0 |
FaceLandmarks68Detector | 1 | 1 | 3.2 | 1226.0 |
FaceLandmarks68Detector | 8 | 1 | 2.0 | 1230.0 |
FaceLandmarks68Detector | 8 | 8 | 1.0 | 1237.0 |
CPU. HumanDetector performance#
The table below shows the performance of HumanDetector on the CPU.
Measurement | CPU threads | BatchSize | Average (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HumanDetector (resize to 320) | 1 | 1 | 10.05 | 1740.0 |
HumanDetector (resize to 320) | 8 | 1 | 6.18 | 1813.0 |
HumanDetector (resize to 320) | 8 | 8 | 3.53 | 1978.0 |
HumanDetector (resize to 640) | 1 | 1 | 35.03 | 1776.0 |
HumanDetector (resize to 640) | 8 | 1 | 14.71 | 1865.0 |
HumanDetector (resize to 640) | 8 | 8 | 11.55 | 2234.0 |
HumanRedetect | 1 | 1 | 2.61 | 1239.0 |
HumanRedetect | 8 | 1 | 2.76 | 1545.0 |
HumanRedetect | 8 | 4 | 1.24 | 1770.0 |
HumanRedetect | 8 | 8 | 1.26 | 1987.0 |
CPU. HumanFaceDetector performance#
The table below shows the performance of HumanFaceDetector on the CPU.
Measurement | CPU threads | BatchSize | Average (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HumanFaceDetector (minFaceSize=20) | 1 | 1 | 425.37 | 2558 |
HumanFaceDetector (minFaceSize=20) | 8 | 1 | 183.5 | 2600 |
HumanFaceDetector (minFaceSize=20) | 8 | 8 | 182.35 | 9340 |
HumanFaceDetector (minFaceSize=50) | 1 | 1 | 66.97 | 1783 |
HumanFaceDetector (minFaceSize=50) | 8 | 1 | 28.9 | 1812 |
HumanFaceDetector (minFaceSize=50) | 8 | 8 | 29.17 | 2900 |
HumanFaceDetector (minFaceSize=90) | 1 | 1 | 22.6 | 1734 |
HumanFaceDetector (minFaceSize=90) | 8 | 1 | 10.71 | 1758 |
HumanFaceDetector (minFaceSize=90) | 8 | 8 | 9.17 | 2072 |
CPU. HeadDetector performance#
Type | CPU threads | Batch Size | Average (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HeadDetector (minHeadSize=20) | 1 | 1 | 322.93 | 2156.0 |
HeadDetector (minHeadSize=20) | 8 | 1 | 118.41 | 2223.0 |
HeadDetector (minHeadSize=20) | 8 | 8 | 109.41 | 5578.0 |
HeadDetector (minHeadSize=50) | 1 | 1 | 57.97 | 1781.0 |
HeadDetector (minHeadSize=50) | 8 | 1 | 23.99 | 1823.0 |
HeadDetector (minHeadSize=50) | 8 | 8 | 19.94 | 2485.0 |
HeadDetector (minHeadSize=90) | 1 | 1 | 23.37 | 1708.0 |
HeadDetector (minHeadSize=90) | 8 | 1 | 10.9 | 1779.0 |
HeadDetector (minHeadSize=90) | 8 | 8 | 7.32 | 2036.0 |
CPU. Estimations performance with batch interface#
The table below shows the performance of Estimations on the CPU for estimators that have a batch interface. All these measurements are performed with minFaceSize=50
.
Measurement | CPU threads | BatchSize | Average (ms) | RAM Memory (Mb) |
---|---|---|---|---|
Eyes (INFRA_RED, useStatusPlan=0) | 1 | 1 | 0.76 | 1525.0 |
Eyes (INFRA_RED, useStatusPlan=0) | 8 | 1 | 0.35 | 1530.0 |
Eyes (INFRA_RED, useStatusPlan=0) | 8 | 8 | 0.19 | 1529.0 |
Eyes (RGB, useStatusPlan=0) | 1 | 1 | 1.22 | 1599.0 |
Eyes (RGB, useStatusPlan=0) | 8 | 1 | 0.42 | 1629.0 |
Eyes (RGB, useStatusPlan=0) | 8 | 8 | 0.24 | 1626.0 |
Eyes (INFRA_RED, useStatusPlan=1) | 1 | 1 | 0.76 | 1524.0 |
Eyes (INFRA_RED, useStatusPlan=1) | 8 | 1 | 0.34 | 1538.0 |
Eyes (INFRA_RED, useStatusPlan=1) | 8 | 8 | 0.19 | 1532.0 |
Eyes (RGB, useStatusPlan=1) | 1 | 1 | 1.21 | 1601.0 |
Eyes (RGB, useStatusPlan=1) | 8 | 1 | 0.4 | 1626.0 |
Eyes (RGB, useStatusPlan=1) | 8 | 8 | 0.24 | 1634.0 |
Infra-Red | 1 | 1 | 2 | 1191.0 |
Infra-Red | 8 | 1 | 1.0 | 1209.0 |
Infra-Red | 8 | 8 | 0.7 | 1218.0 |
AGS | 1 | 1 | 0.24 | 1735.0 |
AGS | 8 | 1 | 0.15 | 1763.0 |
AGS | 8 | 8 | 0.08 | 1804.0 |
HeadPoseByImage | 1 | 1 | 0.24 | 1648.0 |
HeadPoseByImage | 8 | 1 | 0.15 | 1672.0 |
HeadPoseByImage | 8 | 8 | 0.06 | 1712.0 |
Warper | 1 | 1 | 2.1 | 1180.0 |
Warper | 8 | 1 | 2.2 | 1219.0 |
Warper | 8 | 8 | 0.9 | 1230.0 |
BlackWhite | 1 | 1 | 1.3 | 1249.0 |
BlackWhite | 8 | 1 | 0.7 | 1265.0 |
BlackWhite | 8 | 8 | 1.2 | 1263.0 |
BestShotQuality | 1 | 1 | 0.5 | 1833.0 |
BestShotQuality | 8 | 1 | 0.22 | 1857.0 |
BestShotQuality | 8 | 8 | 0.1 | 1896.0 |
MedicalMask | 1 | 1 | 5.6 | 1258.0 |
MedicalMask | 8 | 1 | 3.2 | 1287.0 |
MedicalMask | 8 | 8 | 2.8 | 1318.0 |
LivenessOneShotRGBEstimator | 1 | 1 | 199.57 | 2119.0 |
LivenessOneShotRGBEstimator | 8 | 1 | 51.62 | 2204.0 |
LivenessOneShotRGBEstimator | 8 | 8 | 47.39 | 2570.0 |
Orientation | 1 | 1 | 5.06 | 1609.0 |
Orientation | 8 | 1 | 3.33 | 1682.0 |
Orientation | 8 | 8 | 1.86 | 1875.0 |
CredibilityCheck | 1 | 1 | 120.3 | 1332.0 |
CredibilityCheck | 8 | 1 | 35.1 | 1351.0 |
CredibilityCheck | 8 | 8 | 34.1 | 1558.0 |
FacialHair | 1 | 1 | 12.86 | 1751.0 |
FacialHair | 8 | 1 | 4.84 | 1770.0 |
FacialHair | 8 | 8 | 4.24 | 1794.0 |
PortraitStyle | 1 | 1 | 1.54 | 1738.0 |
PortraitStyle | 8 | 1 | 2.2 | 1846.0 |
PortraitStyle | 8 | 8 | 0.95 | 1915.0 |
Background | 1 | 1 | 1.1 | 1239.0 |
Background | 8 | 1 | 1.2 | 1258.0 |
Background | 8 | 8 | 1.7 | 1305.0 |
NaturalLight | 1 | 1 | 2.37 | 1250.0 |
NaturalLight | 8 | 1 | 1.49 | 1267.0 |
NaturalLight | 8 | 8 | 1.97 | 1276.0 |
FishEye | 1 | 1 | 12.8 | 1747.0 |
FishEye | 8 | 1 | 4.8 | 1747.0 |
FishEye | 8 | 8 | 0.6 | 1771.0 |
RedEye | 1 | 1 | 5.7 | 1241.0 |
RedEye | 8 | 1 | 1.9 | 1260.0 |
RedEye | 8 | 8 | 1.6 | 1264.0 |
HeadWear | 1 | 1 | 4.09 | 1742.0 |
HeadWear | 8 | 1 | 2.63 | 1769.0 |
HeadWear | 8 | 8 | 1.2 | 1773.0 |
EyeBrowEstimator | 1 | 1 | 13.06 | 1751.0 |
EyeBrowEstimator | 8 | 1 | 4.82 | 1769.0 |
EyeBrowEstimator | 8 | 8 | 4.27 | 1781.0 |
HumanAttributeEstimator | 1 | 1 | 11.93 | 1624.0 |
HumanAttributeEstimator | 8 | 1 | 5.83 | 1651.0 |
HumanAttributeEstimator | 8 | 8 | 3.78 | 1699.0 |
Mouth | 1 | 1 | 6.64 | 1252.0 |
Mouth | 8 | 1 | 2.64 | 1271.0 |
Mouth | 8 | 8 | 2.12 | 1290.0 |
CrowdEstimator (Single, minHeadSize=6) | 1 | 1 | 3157.74 | 2613.0 |
CrowdEstimator (Single, minHeadSize=6) | 8 | 1 | 900.79 | 2631.0 |
CrowdEstimator (Single, minHeadSize=6) | 8 | 8 | 615.48 | 8676.0 |
CrowdEstimator (Single, minHeadSize=12) | 1 | 1 | 801.6 | 1969.0 |
CrowdEstimator (Single, minHeadSize=12) | 8 | 1 | 231.88 | 1990.0 |
CrowdEstimator (Single, minHeadSize=12) | 8 | 8 | 147.72 | 3535.0 |
CrowdEstimator (TwoNets, minHeadSize=6) | 1 | 1 | 3085.82 | 2641.0 |
CrowdEstimator (TwoNets, minHeadSize=6) | 8 | 1 | 906.33 | 2714.0 |
CrowdEstimator (TwoNets, minHeadSize=6) | 8 | 8 | 613.95 | 9073.0 |
CrowdEstimator (TwoNets, minHeadSize=12) | 1 | 1 | 819.59 | 2005.0 |
CrowdEstimator (TwoNets, minHeadSize=12) | 8 | 1 | 239.66 | 2072.0 |
CrowdEstimator (TwoNets, minHeadSize=12) | 8 | 8 | 162.99 | 3955.0 |
DynamicRange | 1 | 1 | 1.49 | 1721.0 |
DynamicRange | 8 | 1 | 1.61 | 1749.0 |
DynamicRange | 8 | 8 | 0.81 | 1793.0 |
LivenessDepthRGB | 1 | 1 | 8.06 | 1757.0 |
LivenessDepthRGB | 8 | 1 | 4.13 | 1796.0 |
LivenessDepthRGB | 8 | 8 | 2.96 | 1839.0 |
Glasses | 1 | 1 | 0.86 | 1743.0 |
Glasses | 8 | 1 | 1.01 | 1768.0 |
Glasses | 8 | 8 | 0.42 | 1768.0 |
DeepFake | 1 | 1 | 232.14 | 1808.0 |
DeepFake | 8 | 1 | 70.03 | 1922.0 |
DeepFake | 8 | 8 | 80.65 | 2443.0 |
NIRLivenessEstimator | 1 | 1 | 17.63 | 1549.0 |
NIRLivenessEstimator | 8 | 1 | 12.06 | 1562.0 |
NIRLivenessEstimator | 8 | 8 | 10.71 | 1664.0 |
LivenessRGBMEstimator | 1 | 1 | 29.1 | 1968.0 |
LivenessRGBMEstimator | 8 | 1 | 10.71 | 2037.0 |
LivenessRGBMEstimator | 8 | 8 | 8.74 | 2356.0 |
DepthLivenessEstimator | 1 | 1 | 2.15 | 1856.0 |
DepthLivenessEstimator | 8 | 1 | 1.35 | 1876.0 |
DepthLivenessEstimator | 8 | 8 | 0.84 | 1894.0 |
Attributes | 1 | 1 | 68.89 | 1994.0 |
Attributes | 8 | 1 | 24.7 | 2023.0 |
Attributes | 8 | 8 | 19.32 | 2274.0 |
FaceOcclusionBatch | 1 | 1 | 7.35 | 1303.0 |
FaceOcclusionBatch | 1 | 8 | 3.61 | 1469.0 |
FaceOcclusionBatch | 8 | 8 | 3.03 | 1455.0 |
CPU. Estimations performance without batch interface#
The table below shows the performance of Estimations on the CPU for estimators that do not have a batch interface. All these measurements are performed with minFaceSize=50
.
Measurement | CPU threads | Average (ms) | RAM Memory (Mb) |
---|---|---|---|
EyesGaze | 1 | 2.2 | 1250 |
EyesGaze | 8 | 1.4 | 1270 |
Emotions | 1 | 13.6 | 1262 |
Emotions | 8 | 4.9 | 1275 |
Quality | 1 | 1.2 | 1178 |
Quality | 8 | 0.6 | 1220 |
Overlap | 1 | 4.5 | 1248 |
Overlap | 8 | 1.3 | 1267 |
PPE | 1 | 11.74 | 1711.0 |
PPE | 8 | 5.6 | 1733.0 |
LivenessFlyingFaces | 1 | 15.07 | 1804 |
LivenessFlyingFaces | 8 | 7.21 | 1913 |
LivenessFPR | 1 | 44.2 | 1263 |
LivenessFPR | 8 | 19.9 | 1293 |
Fights | 1 | 250.26 | 1876 |
Fights | 8 | 63.9 | 1895 |
CPU. Extractor performance#
The table below shows the performance of Extractor on the CPU.
Model | CPU threads | Batch Size | Average (ms) | RAM Memory (Mb) |
---|---|---|---|---|
58 | 1 | 1 | 219.3 | 1470 |
58 | 8 | 8 | 58.0 | 1543 |
59 | 1 | 1 | 219.7 | 1473 |
59 | 8 | 8 | 58.2 | 1550 |
60 | 1 | 1 | 258.0 | 1473 |
60 | 8 | 8 | 51.1 | 1550 |
62 | 1 | 1 | 254.36 | 2007 |
62 | 8 | 1 | 67.54 | 2008 |
62 | 8 | 8 | 71.48 | 2025 |
65 | 1 | 1 | 364.93 | 1992 |
65 | 8 | 1 | 120.88 | 1993 |
65 | 8 | 8 | 93.0 | 2616 |
105 | 1 | 1 | 1.66 | 1604 |
105 | 8 | 8 | 0.71 | 1657 |
106 | 1 | 1 | 140.76 | 1892 |
106 | 8 | 8 | 39.01 | 1954 |
107 | 1 | 1 | 12.0 | 1637 |
107 | 8 | 8 | 3.7 | 1723 |
108 | 1 | 1 | 1.69 | 1606 |
108 | 8 | 8 | 0.72 | 1671 |
109 | 1 | 1 | 133.7 | 1822 |
109 | 8 | 8 | 37.33 | 1889 |
110 | 1 | 1 | 15.53 | 1640 |
110 | 8 | 8 | 5.39 | 1733 |
112 | 1 | 1 | 112.33 | 1823.0 |
112 | 8 | 1 | 39.73 | 1839.0 |
112 | 8 | 8 | 32.95 | 1884.0 |
113 | 1 | 1 | 15.17 | 1640.0 |
113 | 8 | 1 | 6.57 | 1656.0 |
113 | 8 | 8 | 4.7 | 1727.0 |
115 | 1 | 1 | 117.12 | 1920.0 |
115 | 8 | 1 | 41.21 | 1947.0 |
115 | 8 | 8 | 33.19 | 1967.0 |
116 | 1 | 1 | 16.79 | 1739.0 |
116 | 8 | 1 | 7.23 | 1759.0 |
116 | 8 | 8 | 5.07 | 1811.0 |
CPU. Matcher performance#
The table below shows the performance of Matcher on the CPU. The table includes average matcher per second for descriptors received using the following CNN model versions:
- face descriptors: 59, 60, 62
- human body descriptors: 105, 106, 107, 108, 109, 110, 112, 113, 115, 116
Model | CPU threads | Batch Size | Average (matches/sec) | RAM Memory (Mb) |
---|---|---|---|---|
58 | 1 | 1000 | 28 M | 15.0 |
59 | 1 | 1000 | 28 M | 15.0 |
60 | 1 | 1000 | 28 M | 15.0 |
62 | 1 | 1000 | 28 M | 15.0 |
65 | 1 | 1000 | 28 M | 15.0 |
105 | 1 | 1000 | 27.78 M | 113 |
106 | 1 | 1000 | 28.67 M | 112 |
107 | 1 | 1000 | 27.34 M | 113 |
108 | 1 | 1000 | 31.89 M | 117 |
109 | 1 | 1000 | 29.23 M | 114 |
110 | 1 | 1000 | 27.41 M | 112 |
112 | 1 | 1000 | 30 M | 109.0 |
113 | 1 | 1000 | 28.32 | 112.0 |
115 | 1 | 1000 | 31.6 | 112.0 |
116 | 1 | 1000 | 28.7 | 112.0 |
Note: The above value is the maximum performance of the matcher on a particular piece of hardware. Performance in general does not depend on the size of the batch, but may be limited by memory performance at large values of the batch size.
GPU performance#
Benchmarking for GPU was performed on the following hardware configuration:
GPU: NVIDIA Tesla T4.
OS: CentOS Linux release 8.3.2011
GPU. Detector performance#
The table below shows the performance of FaceDetV3 Detector on the GPU.
Measurement | Batch Size | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|---|
Detector (minFaceSize=20) | 1 | 29.02 | 1485.0 | 1663.0 |
Detector (minFaceSize=20) | 4 | 34.37 | 3611.0 | 1691.0 |
Detector (minFaceSize=20) | 8 | 38.09 | 6539.0 | 1741.0 |
Detector (minFaceSize=50) | 1 | 7.46 | 847.0 | 1653.0 |
Detector (minFaceSize=50) | 4 | 6.56 | 1207.0 | 1682.0 |
Detector (minFaceSize=50) | 8 | 6.24 | 1779.0 | 1702.0 |
Detector (minFaceSize=90) | 1 | 4.95 | 835.0 | 1655.0 |
Detector (minFaceSize=90) | 4 | 3.44 | 907.0 | 1669.0 |
Detector (minFaceSize=90) | 8 | 3.17 | 1381.0 | 1694.0 |
Redetect | 1 | 2.52 | 847.0 | 1657.0 |
Redetect | 4 | 1.64 | 1207.0 | 1660.0 |
Redetect | 8 | 1.47 | 1779.0 | 1663.0 |
Redetect | 16 | 1.38 | 2781.0 | 1667.0 |
FaceLandmarks5Detector | 1 | 2.33 | 821.0 | 1651.0 |
FaceLandmarks5Detector | 8 | 0.32 | 821.0 | 1651.0 |
FaceLandmarks5Detector | 16 | 0.17 | 821.0 | 1657.0 |
FaceLandmarks68Detector | 1 | 2.6 | 821.0 | 1669.0 |
FaceLandmarks68Detector | 8 | 1.5 | 821.0 | 1668.3 |
FaceLandmarks68Detector | 16 | 1.4 | 949.0 | 1663.0 |
GPU. HumanDetector performance#
The table below shows the performance of HumanDetector on the GPU.
Measurement | Batch Size | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|---|
HumanDetector (resize to 320) | 1 | 4.17 | 779.0 | 1778.0 |
HumanDetector (resize to 320) | 4 | 2.46 | 819.0 | 1792.0 |
HumanDetector (resize to 320) | 8 | 2.17 | 909.0 | 1815.0 |
HumanDetector (resize to 640) | 1 | 5.42 | 827.0 | 1784.0 |
HumanDetector (resize to 640) | 4 | 4.14 | 1013.0 | 1796.0 |
HumanDetector (resize to 640) | 8 | 3.92 | 1371.0 | 1824.0 |
HumanRedetect | 1 | 2.74 | 789.0 | 1696.0 |
HumanRedetect | 4 | 1.67 | 1013.0 | 1695.0 |
HumanRedetect | 8 | 1.47 | 1251.0 | 1689.0 |
HumanRedetect | 16 | 1.4 | 1867.0 | 1709.0 |
GPU. HeadDetector performance#
The table below shows the performance of HeadDetector on the GPU.
Type | Batch Size | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|---|
HeadDetector (minHeadSize=20) | 1 | 24.38 | 1561.0 | 1730.0 |
HeadDetector (minHeadSize=20) | 4 | 31.35 | 4103.0 | 1745.0 |
HeadDetector (minHeadSize=20) | 8 | 35.85 | 7491.0 | 1799.0 |
HeadDetector (minHeadSize=50) | 1 | 6.63 | 837.0 | 1716.0 |
HeadDetector (minHeadSize=50) | 4 | 5.74 | 1367.0 | 1749.0 |
HeadDetector (minHeadSize=50) | 8 | 5.45 | 1931.0 | 1767.0 |
HeadDetector (minHeadSize=90) | 1 | 4.41 | 749.0 | 1720.0 |
HeadDetector (minHeadSize=90) | 4 | 3.04 | 905.0 | 1734.0 |
HeadDetector (minHeadSize=90) | 8 | 2.8 | 1103.0 | 1759.0 |
GPU. HumanFace detector performance#
The table below shows the performance of HumanFaceDetector on the GPU.
Measurement | Batch Size | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|---|
HumanFaceDetector (minFaceSize=20) | 1 | 34.1 | 1675.0 | 1703.0 |
HumanFaceDetector (minFaceSize=20) | 4 | 42.6 | 4415.0 | 1774.0 |
HumanFaceDetector (minFaceSize=20) | 8 | 50.32 | 8041.0 | 1889.0 |
HumanFaceDetector (minFaceSize=50) | 1 | 7.99 | 903.0 | 1674.0 |
HumanFaceDetector (minFaceSize=50) | 4 | 7.15 | 1487.0 | 1706.0 |
HumanFaceDetector (minFaceSize=50) | 8 | 6.83 | 2067.0 | 1764.0 |
HumanFaceDetector (minFaceSize=90) | 1 | 5.3 | 903.0 | 1672.0 |
HumanFaceDetector (minFaceSize=90) | 4 | 3.52 | 929.0 | 1685.0 |
HumanFaceDetector (minFaceSize=90) | 8 | 3.24 | 1125.0 | 1719.0 |
GPU. Estimations performance with batch interface#
The table below shows the performance of Estimations on the GPU for estimators that have a batch interface. All these measurements are performed with minFaceSize=50
.
Measurement | Batch Size | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|---|
HeadPoseByImage | 1 | 2.26 | 785.0 | 1692.0 |
HeadPoseByImage | 16 | 1.45 | 881.0 | 1775.0 |
HeadPoseByImage | 32 | 1.42 | 975.0 | 1873.0 |
Warper | 1 | 0.11 | 739.0 | 1672.0 |
Warper | 32 | 0.03 | 931.0 | 1672.0 |
Eyes (INFRA_RED, useStatusPlan=0) | 1 | 0.51 | 600.0 | 1654.0 |
Eyes (INFRA_RED, useStatusPlan=0) | 16 | 0.11 | 600.0 | 1652.0 |
Eyes (INFRA_RED, useStatusPlan=0) | 32 | 0.09 | 632.0 | 1654.0 |
Eyes (RGB, useStatusPlan=0) | 1 | 0.82 | 712.0 | 1659.0 |
Eyes (RGB, useStatusPlan=0) | 16 | 0.11 | 744.0 | 1656.0 |
Eyes (RGB, useStatusPlan=0) | 32 | 0.1 | 744.0 | 1655.0 |
Eyes (INFRA_RED, useStatusPlan=1) | 1 | 0.55 | 600.0 | 1653.0 |
Eyes (INFRA_RED, useStatusPlan=1) | 16 | 0.11 | 600.0 | 1645.0 |
Eyes (INFRA_RED, useStatusPlan=1) | 32 | 0.1 | 632.0 | 1648.0 |
Eyes (RGB, useStatusPlan=1) | 1 | 0.88 | 712.0 | 1663.0 |
Eyes (RGB, useStatusPlan=1) | 16 | 0.12 | 744.0 | 1662.0 |
Eyes (RGB, useStatusPlan=1) | 32 | 0.11 | 744.0 | 1667.0 |
Infra-Red | 1 | 1.11 | 811.0 | 1666.0 |
Infra-Red | 32 | 0.54 | 811.0 | 1679.0 |
AGS | 1 | 2.28 | 899.0 | 1689.0 |
AGS | 16 | 1.42 | 899.0 | 1777.0 |
AGS | 32 | 1.39 | 1089.0 | 1874.0 |
BlackWhite | 1 | 1.05 | 821.0 | 1676.0 |
BlackWhite | 16 | 0.4 | 853.0 | 1677.0 |
BestShotQuality | 1 | 3.11 | 855.0 | 1821.0 |
BestShotQuality | 16 | 1.44 | 855.0 | 1914.0 |
BestShotQuality | 32 | 1.41 | 1045.0 | 2008.0 |
MedicalMask | 1 | 5.01 | 821.0 | 1702.0 |
MedicalMask | 16 | 1.69 | 917.0 | 1791.0 |
LivenessOneShotRGBEstimator | 1 | 13.44 | 1046.0 | 2091.0 |
LivenessOneShotRGBEstimator | 8 | 10.61 | 1614.0 | 2092.0 |
LivenessOneShotRGBEstimator | 16 | 10.3 | 2062.0 | 2091.0 |
Orientation | 1 | 3.12 | 799.0 | 1670.0 |
Orientation | 16 | 1.73 | 963.0 | 1664.0 |
Orientation | 32 | 1.69 | 1141.0 | 1669.0 |
CredibilityCheck | 1 | 5.54 | 947.0 | 1774.0 |
CredibilityCheck | 16 | 3.72 | 1339.0 | 1771.0 |
FacialHair | 1 | 1.86 | 853.0 | 1687.0 |
FacialHair | 16 | 0.32 | 853.0 | 1683.0 |
FacialHair | 32 | 0.28 | 853.0 | 1685.0 |
PortraitStyle | 1 | 2.84 | 895.0 | 1671.0 |
PortraitStyle | 16 | 1.51 | 915.0 | 1770.0 |
PortraitStyle | 32 | 1.48 | 1085.0 | 1861.0 |
Background | 1 | 2.6 | 821.0 | 1679.0 |
Background | 16 | 1.5 | 917.0 | 1770.0 |
NaturalLight | 1 | 3.61 | 853.0 | 1692.0 |
NaturalLight | 16 | 0.27 | 853.0 | 1695.0 |
FishEye | 1 | 2.37 | 895.0 | 1692.0 |
FishEye | 16 | 0.14 | 895.0 | 1694.0 |
RedEye | 1 | 1.1 | 821.0 | 1675.0 |
RedEye | 16 | 0.15 | 821.0 | 1675.0 |
HeadWear | 1 | 4.14 | 853.0 | 1696.0 |
HeadWear | 16 | 0.36 | 853.0 | 1699.0 |
HeadWear | 32 | 0.27 | 853.0 | 1697.0 |
EyeBrowEstimator | 1 | 2.56 | 895.0 | 1694.0 |
EyeBrowEstimator | 16 | 0.8 | 895.0 | 1693.0 |
EyeBrowEstimator | 32 | 0.76 | 803.0 | 1079.0 |
HumanAttributeEstimator | 1 | 5.53 | 853.0 | 1691.0 |
HumanAttributeEstimator | 16 | 0.57 | 853.0 | 1722.0 |
Mouth | 1 | 4.03 | 853.0 | 1690.0 |
Mouth | 16 | 0.42 | 949.0 | 1691.0 |
Mouth | 32 | 0.37 | 1043.0 | 1690.0 |
Glasses | 1 | 1.41 | 901.0 | 1695.0 |
Glasses | 16 | 0.2 | 901.0 | 1689.0 |
Glasses | 32 | 0.16 | 901.0 | 1686.0 |
CrowdEstimator (Single, minHeadSize=6) | 1 | 64.57 | 1569.0 | 1843.0 |
CrowdEstimator (Single, minHeadSize=6) | 4 | 65.7 | 3185.0 | 1873.0 |
CrowdEstimator (Single, minHeadSize=6) | 8 | 66.96 | 3334.0 | 1904.0 |
CrowdEstimator (Single, minHeadSize=12) | 1 | 22.15 | 985.0 | 1834.0 |
CrowdEstimator (Single, minHeadSize=12) | 4 | 21.38 | 1433.0 | 1857.0 |
CrowdEstimator (Single, minHeadSize=12) | 8 | 21.67 | 1496.0 | 1883.0 |
CrowdEstimator (TwoNets, minHeadSize=6) | 1 | 69.7 | 1745.0 | 1854.0 |
CrowdEstimator (TwoNets, minHeadSize=6) | 4 | 71.11 | 3570.0 | 1903.0 |
CrowdEstimator (TwoNets, minHeadSize=6) | 8 | 72.04 | 4164.0 | 1925.0 |
CrowdEstimator (TwoNets, minHeadSize=12) | 1 | 26.89 | 1083.0 | 1846.0 |
CrowdEstimator (TwoNets, minHeadSize=12) | 4 | 23.8 | 1770.0 | 1871.0 |
CrowdEstimator (TwoNets, minHeadSize=12) | 8 | 25.44 | 2208.0 | 1904.0 |
DeepFake | 1 | 14.48 | 0.0 | 1807.0 |
DeepFake | 16 | 13.09 | 0.0 | 1914.0 |
DeepFake | 32 | 13.15 | 0.0 | 1985.0 |
LivenessDepthRGB | 1 | 4.79 | 931.0 | 1717.0 |
LivenessDepthRGB | 16 | 3.91 | 975.0 | 1809.0 |
LivenessDepthRGB | 32 | 3.9 | 1127.0 | 1914.0 |
NIRLivenessEstimator | 1 | 8.99 | 610.0 | 1659.0 |
NIRLivenessEstimator | 16 | 8.15 | 708.0 | 1757.0 |
NIRLivenessEstimator | 32 | 8.14 | 836.0 | 1852.0 |
LivenessRGBMEstimator | 1 | 6.56 | 871.0 | 1938.0 |
LivenessRGBMEstimator | 16 | 4.18 | 1625.0 | 2085.0 |
LivenessRGBMEstimator | 32 | 4.9 | 2225.0 | 2238.0 |
DepthLivenessEstimator | 1 | 2.08 | 737.0 | 1927.0 |
DepthLivenessEstimator | 16 | 0.44 | 771.0 | 1932.0 |
DepthLivenessEstimator | 32 | 0.38 | 805.0 | 1936.0 |
Attributes | 1 | 3.75 | 871.0 | 1984.0 |
Attributes | 16 | 1.97 | 1373.0 | 1980.0 |
Attributes | 32 | 1.9 | 1895.0 | 1991.0 |
FaceOcclusionBatch | 1 | 1.64 | 620.0 | 1281.0 |
FaceOcclusionBatch | 16 | 0.76 | 844.0 | 1330.0 |
FaceOcclusionBatch | 32 | 0.73 | 1036.0 | 1324.0 |
GPU. Estimations performance without batch interface#
The table below shows the performance of Estimations on the GPU for estimators that do not have a batch interface. All these measurements are performed with minFaceSize=50
.
Measurement | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|
EyesGaze | 1.65 | 821 | 1675 |
Emotions | 1.99 | 821 | 1689 |
Quality | 0.98 | 731 | 1665 |
Overlap | 1.23 | 821 | 1688 |
PPE | 3.32 | 803.0 | 1718.0 |
LivenessFlyingFaces | 6.39 | 927 | 1694 |
LivenessFPR | 12.56 | 885 | 1697 |
Fights | 14.56 | 1093 | 1874 |
GPU. Extractor performance#
The table below shows the performance of Extractor on the GPU.
Model | Batch Size | Average (ms) | GPU Memory (Mb) | RAM Memory (Mb) |
---|---|---|---|---|
58 | 1 | 10.2 | 989.0 | 1835 |
58 | 16 | 6.4 | 1781.0 | 1825 |
59 | 1 | 10.2 | 929.0 | 1833 |
59 | 16 | 6.4 | 1341.0 | 1837 |
60 | 1 | 16.0 | 931.0 | 1840 |
60 | 16 | 8.9 | 1343.0 | 1845 |
62 | 1 | 11.23 | 1043.0 | 2009.0 |
62 | 8 | 7.81 | 1227.0 | 2006.0 |
62 | 16 | 7.75 | 1437.0 | 2016.0 |
65 | 1 | 6.48 | 949.0 | 1995 |
65 | 8 | 3.47 | 1911.0 | 1996 |
65 | 16 | 3.34 | 2439.0 | 1996 |
105 | 1 | 3.48 | 785 | 1664 |
105 | 16 | 0.3 | 815 | 1673 |
106 | 1 | 6.28 | 973 | 1893 |
106 | 16 | 9.38 | 1371 | 1894 |
107 | 1 | 3.41 | 807 | 1698 |
107 | 16 | 0.59 | 911 | 1696 |
108 | 1 | 3.47 | 785 | 1654 |
108 | 16 | 0.3 | 815 | 1672 |
109 | 1 | 6.22 | 933 | 1833 |
109 | 16 | 7.83 | 1261 | 1833 |
110 | 1 | 3.38 | 809 | 1693 |
110 | 16 | 0.76 | 939 | 1693 |
112 | 1 | 6.52 | 901.0 | 1836.0 |
112 | 8 | 3.71 | 1029.0 | 1834.0 |
112 | 16 | 3.57 | 1209.0 | 1835.0 |
113 | 1 | 3.13 | 809.0 | 1696.0 |
113 | 8 | 0.82 | 873.0 | 1697.0 |
113 | 16 | 0.68 | 937.0 | 1703.0 |
115 | 1 | 6.56 | 877.0 | 1925.0 |
115 | 8 | 5.51 | 1001.0 | 1931.0 |
115 | 16 | 5.43 | 1141.0 | 1932.0 |
116 | 1 | 2.92 | 753.0 | 1783.0 |
116 | 8 | 0.85 | 819.0 | 1804.0 |
116 | 16 | 0.73 | 885.0 | 1804.0 |
NPU Performance#
Benchmarking for NPU was performed on the server with the following hardware configuration:
NPU: Huawei Atlas 300I (inference card).
OS: Ubuntu 18.04
CPU: Intel(R) Xeon(R) Gold 5118 CPU @ 2.30GHz x 48
RAM: 64GB
NPU. Detector performance#
The table below shows the performance of Detector on the NPU.
Measurement | BatchSize | Average (ms) |
---|---|---|
Detector (minFaceSize=20) | 1 | 24.4 |
Detector (minFaceSize=20) | 4 | 18.01 |
Detector (minFaceSize=20) | 8 | 17.73 |
Detector (minFaceSize=50) | 1 | 24.53 |
Detector (minFaceSize=50) | 4 | 18.0 |
Detector (minFaceSize=50) | 8 | 17.74 |
Detector (minFaceSize=90) | 1 | 24.44 |
Detector (minFaceSize=90) | 4 | 17.91 |
Detector (minFaceSize=90) | 8 | 17.44 |
Redetect | 1 | 7.56 |
Redetect | 8 | 4.31 |
Redetect | 16 | 4.08 |
NPU. Estimations performance with batch interface#
The table below shows the performance of Estimations on the NPU for estimators that have a batch interface. All these measurements are performed with minFaceSize=50
.
Measurement | BatchSize | Average (ms) |
---|---|---|
HeadPoseByImage | 1 | 8.0 |
HeadPoseByImage | 16 | 4.2 |
HeadPoseByImage | 32 | 3.9 |
AGS | 1 | 6.6 |
AGS | 16 | 3.7 |
AGS | 32 | 3.7 |
BestShotQuality | 1 | 15.6 |
BestShotQuality | 16 | 7.8 |
BestShotQuality | 32 | 7.6 |
MedicalMask | 1 | 6.1 |
MedicalMask | 16 | 3.8 |
MedicalMask | 32 | 3.7 |
NPU. Estimations performance without batch interface#
The table below shows the performance of Estimations on the NPU for estimators that do not have a batch interface. All these measurements are performed with minFaceSize=50
.
Measurement | Average (ms) |
---|---|
Warper | 2.1 |
NPU. Extractor performance#
The table below shows the performance of Extractor on the NPU.
Type | Model | Batch Size | Average (ms) |
---|---|---|---|
Extractor | 57 | 1 | 10.9 |
Extractor | 57 | 16 | 7.4 |
Rockchip (Ubuntu 24.04 LTS)#
The number of threads auto
means that will be taken the maximum number of available threads.
For this mode use the -1
value for the numThreads
parameter in the runtime.conf
.
This number of threads is equal to according number of available processor cores.
We strongly recommend you to follow this recommendation; otherwise, performance can be significantly reduced.
Description of according settings you can find in "Configuration Guide - Runtime settings".
The performance measurements are presented for device with configurations as below:
Architecture: aarch64 Byte Order: Little Endian CPU(s): 8 On-line CPU(s) list: 0-7 Thread(s) per core: 1 Core(s) per socket: 4 Socket(s): 1 Vendor ID: ARM Model: 0 Model name: Cortex-A55 Stepping: r2p0 CPU max MHz: 1800.0000 CPU min MHz: 408.0000 BogoMIPS: 48.00 Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm lrcpc dcpop asimddp
The number of threads you can find in tables below.
*Note: In the case of these tests, power
and weak
refer to a Linux command (taskset -c j,k, where j
and k
are CPU cores)
that explicitly sets the CPU affinity of a process. In simple terms, it tells the system to run the process only on the specified CPU cores.
Power
stands for taskset -c 4-7
and weak
stands for taskset -c 0-3
.
Rockchip (power) environment. Detector performance#
The table below shows the performance of Detector on the Rockchip (power) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
Detector (minFaceSize=20) | 1 | 1 | 4048.29 | 604.0 |
Detector (minFaceSize=20) | 2 | 1 | 4218.24 | 601.0 |
Detector (minFaceSize=20) | 2 | 8 | 4148.53 | 4495.0 |
Detector (minFaceSize=50) | 1 | 1 | 548.7 | 132.0 |
Detector (minFaceSize=50) | 2 | 1 | 559.4 | 136.0 |
Detector (minFaceSize=50) | 2 | 8 | 552.05 | 809.0 |
Detector (minFaceSize=90) | 1 | 1 | 157.16 | 71.0 |
Detector (minFaceSize=90) | 2 | 1 | 170.72 | 73.0 |
Detector (minFaceSize=90) | 2 | 8 | 179.77 | 326.0 |
Redetect | 1 | 1 | 3.41 | 126.0 |
Redetect | 2 | 1 | 3.47 | 127.0 |
Redetect | 2 | 8 | 3.26 | 768.0 |
Landmarks5Detector | 1 | 1 | 1.13 | 136.0 |
Landmarks5Detector | 2 | 1 | 1.18 | 137.0 |
Landmarks5Detector | 2 | 8 | 1.15 | 137.0 |
Landmarks68Detector | 1 | 1 | 8.62 | 136.0 |
Landmarks68Detector | 2 | 1 | 8.62 | 137.0 |
Landmarks68Detector | 2 | 8 | 8.82 | 137.0 |
Rockchip (power) environment. Extractor performance#
The table below shows the performance of Extractor on the Rockchip (power) environment.
Model | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
62 | 1 | 1 | 2130.74 | 389.0 |
62 | 2 | 1 | 2110.69 | 387.0 |
62 | 2 | 8 | 2216.14 | 387.0 |
Rockchip (power) environment. HeadDetector performance#
The table below shows the performance of HeadDetector on the Rockchip (power) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HeadDetector (minHeadSize=20) | 1 | 1 | 4153.01 | 599.0 |
HeadDetector (minHeadSize=20) | 2 | 1 | 4261.43 | 596.0 |
HeadDetector (minHeadSize=50) | 1 | 1 | 539.03 | 131.0 |
HeadDetector (minHeadSize=50) | 2 | 1 | 556.37 | 132.0 |
HeadDetector (minHeadSize=50) | 2 | 8 | 550.99 | 809.0 |
HeadDetector (minHeadSize=90) | 1 | 1 | 154.06 | 71.0 |
HeadDetector (minHeadSize=90) | 2 | 1 | 156.39 | 71.0 |
HeadDetector (minHeadSize=90) | 2 | 8 | 179.22 | 324.0 |
Rockchip (power) environment. HumanDetector performance#
The table below shows the performance of HumanDetector on the Rockchip (power) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HumanDetector (resize to 320) | 1 | 1 | 70.32 | 56.0 |
HumanDetector (resize to 320) | 2 | 1 | 80.67 | 55.0 |
HumanDetector (resize to 320) | 2 | 8 | 83.49 | 177.0 |
HumanDetector (resize to 640) | 1 | 1 | 316.34 | 89.0 |
HumanDetector (resize to 640) | 2 | 1 | 321.24 | 90.0 |
HumanDetector (resize to 640) | 2 | 8 | 352.26 | 454.0 |
HumanRedetect | 1 | 1 | 14.49 | 88.0 |
HumanRedetect | 2 | 1 | 14.38 | 88.0 |
HumanRedetect | 2 | 8 | 14.42 | 413.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=20) | 1 | 1 | 4665.39 | 788.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=20) | 2 | 1 | 4757.78 | 788.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=20) | 2 | 8 | 4833.92 | 6015.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=50) | 1 | 1 | 625.86 | 161.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=50) | 2 | 1 | 631.68 | 162.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=50) | 2 | 8 | 631.79 | 1058.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=90) | 1 | 1 | 188.8 | 80.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=90) | 2 | 1 | 181.36 | 80.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=90) | 2 | 8 | 202.62 | 406.0 |
HumanFaceDetectorBoxes (minFaceSize=20) | 1 | 1 | 4532.06 | 735.0 |
HumanFaceDetectorBoxes (minFaceSize=20) | 2 | 1 | 4478.3 | 735.0 |
HumanFaceDetectorBoxes (minFaceSize=20) | 2 | 8 | 4707.63 | 5617.0 |
HumanFaceDetectorBoxes (minFaceSize=50) | 1 | 1 | 593.94 | 153.0 |
HumanFaceDetectorBoxes (minFaceSize=50) | 2 | 1 | 604.22 | 153.0 |
HumanFaceDetectorBoxes (minFaceSize=50) | 2 | 8 | 597.29 | 992.0 |
HumanFaceDetectorBoxes (minFaceSize=90) | 1 | 1 | 187.39 | 77.0 |
HumanFaceDetectorBoxes (minFaceSize=90) | 2 | 1 | 171.93 | 78.0 |
HumanFaceDetectorBoxes (minFaceSize=90) | 2 | 8 | 193.81 | 384.0 |
HumanWarper | 1 | 1 | 0.64 | 51.0 |
HumanWarper | 2 | 1 | 0.59 | 52.0 |
HumanWarper | 2 | 8 | 1.02 | 93.0 |
HumanWarper | 1 | 1 | 0.64 | 86.0 |
HumanWarper | 2 | 1 | 0.61 | 87.0 |
HumanWarper | 2 | 8 | 1.01 | 128.0 |
Rockchip (power) Estimations performance without batch interface#
The table below shows the performance of Estimations on the CPU for estimators that do not have a batch interface.
All these measurements are performed with minFaceSize=50
.
Type | CPU threads | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|
LivenessFPR | 2 | 179.91 | 152.0 |
LivenessFPR | 1 | 350.14 | 150.0 |
PPE | 2 | 38.26 | 100.0 |
PPE | 1 | 69.45 | 100.0 |
Overlap | 2 | 15.91 | 139.0 |
Overlap | 1 | 29.76 | 140.0 |
Rockchip (power) environment. Estimations performance with batch interface#
The table below shows the performance of Estimations on the Rockchip (power) environment for estimators that have a batch interface.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HeadPose | 1 | 1 | 0.92 | 139.0 |
HeadPose | 2 | 1 | 0.91 | 138.0 |
HeadPose | 2 | 8 | 0.99 | 179.0 |
Warper | 1 | 1 | 5.82 | 130.0 |
Warper | 2 | 1 | 5.85 | 130.0 |
Warper | 2 | 8 | 5.84 | 131.0 |
Eyes | 1 | 1 | 5.4 | 132.0 |
Eyes | 2 | 1 | 5.67 | 133.0 |
Eyes | 2 | 8 | 5.94 | 137.0 |
Eyes | 1 | 1 | 5.4 | 132.0 |
Eyes | 2 | 1 | 5.68 | 133.0 |
Eyes | 2 | 8 | 6.39 | 133.0 |
Eyes | 1 | 1 | 2.63 | 48.0 |
Eyes | 2 | 1 | 2.9 | 48.0 |
Eyes | 2 | 8 | 3.02 | 54.0 |
Eyes | 1 | 1 | 2.65 | 48.0 |
Eyes | 2 | 1 | 2.76 | 48.0 |
Eyes | 2 | 8 | 2.91 | 54.0 |
InfraRed | 1 | 1 | 13.86 | 51.0 |
InfraRed | 2 | 1 | 7.78 | 51.0 |
InfraRed | 2 | 8 | 11.29 | 70.0 |
AGS | 1 | 1 | 0.95 | 138.0 |
AGS | 2 | 1 | 0.95 | 138.0 |
AGS | 2 | 8 | 0.96 | 179.0 |
BestShotQuality | 1 | 1 | 1.99 | 139.0 |
BestShotQuality | 2 | 1 | 2.04 | 141.0 |
BestShotQuality | 2 | 8 | 1.92 | 182.0 |
MedicalMask | 1 | 1 | 37.75 | 157.0 |
MedicalMask | 2 | 1 | 38.42 | 158.0 |
MedicalMask | 2 | 8 | 38.2 | 200.0 |
LivenessOneShotRGBEstimator | 1 | 1 | 1518.52 | 271.0 |
LivenessOneShotRGBEstimator | 2 | 1 | 1505.27 | 272.0 |
Orientation | 1 | 1 | 61.14 | 37.0 |
Orientation | 2 | 1 | 50.35 | 40.0 |
Orientation | 2 | 8 | 68.16 | 85.0 |
FacialHair | 1 | 1 | 115.42 | 150.0 |
FacialHair | 2 | 1 | 127.69 | 149.0 |
FacialHair | 2 | 8 | 132.92 | 149.0 |
CredibilityCheck | 1 | 1 | 1088.63 | 224.0 |
CredibilityCheck | 2 | 1 | 1178.26 | 223.0 |
CredibilityCheck | 2 | 8 | 1252.27 | 223.0 |
BlackWhite | 1 | 1 | 7.71 | 136.0 |
BlackWhite | 2 | 1 | 4.33 | 137.0 |
BlackWhite | 2 | 8 | 4.46 | 138.0 |
NaturalLight | 1 | 1 | 15.15 | 140.0 |
NaturalLight | 2 | 1 | 15.7 | 140.0 |
NaturalLight | 2 | 8 | 14.56 | 140.0 |
PortraitStyle | 1 | 1 | 6.77 | 138.0 |
PortraitStyle | 2 | 1 | 7.17 | 139.0 |
PortraitStyle | 2 | 8 | 7.62 | 180.0 |
FishEye | 1 | 1 | 16.59 | 141.0 |
FishEye | 2 | 1 | 18.1 | 143.0 |
FishEye | 2 | 8 | 20.93 | 143.0 |
EyeBrow | 1 | 1 | 116.83 | 150.0 |
EyeBrow | 2 | 1 | 114.04 | 149.0 |
EyeBrow | 2 | 8 | 132.6 | 149.0 |
HumanAttribute | 1 | 1 | 94.86 | 59.0 |
HumanAttribute | 2 | 1 | 97.65 | 59.0 |
HumanAttribute | 2 | 8 | 96.34 | 75.0 |
RedEye | 1 | 1 | 17.87 | 134.0 |
RedEye | 2 | 1 | 30.42 | 135.0 |
RedEye | 2 | 8 | 19.22 | 135.0 |
HeadWear | 1 | 1 | 26.8 | 150.0 |
HeadWear | 2 | 1 | 27.28 | 149.0 |
HeadWear | 2 | 8 | 24.6 | 149.0 |
Background | 1 | 1 | 6.61 | 138.0 |
Background | 2 | 1 | 7.05 | 139.0 |
Background | 2 | 8 | 7.54 | 180.0 |
Mouth | 1 | 1 | 51.22 | 141.0 |
Mouth | 2 | 1 | 52.67 | 141.0 |
Mouth | 2 | 8 | 60.71 | 141.0 |
Attributes | 1 | 1 | 590.98 | 182.0 |
Attributes | 2 | 1 | 531.96 | 182.0 |
Attributes | 2 | 8 | 552.64 | 274.0 |
Quality | 1 | 1 | 6.28 | 133.0 |
Quality | 2 | 1 | 6.33 | 132.0 |
Quality | 2 | 8 | 7.44 | 132.0 |
Emotions | 1 | 1 | 114.09 | 149.0 |
Emotions | 2 | 1 | 116.42 | 149.0 |
Emotions | 2 | 8 | 135.98 | 149.0 |
EyesGaze | 1 | 1 | 16.11 | 137.0 |
EyesGaze | 2 | 1 | 9.09 | 137.0 |
EyesGaze | 2 | 8 | 9.97 | 139.0 |
Glasses | 1 | 1 | 6.33 | 133.0 |
Glasses | 2 | 1 | 6.85 | 133.0 |
Glasses | 2 | 8 | 7.33 | 133.0 |
LivenessFlyingFaces | 1 | 1 | 86.67 | 155.0 |
LivenessFlyingFaces | 2 | 1 | 92.49 | 158.0 |
LivenessFlyingFaces | 2 | 8 | 100.76 | 197.0 |
DynamicRange | 1 | 1 | 0.52 | 135.0 |
DynamicRange | 2 | 1 | 0.49 | 136.0 |
DynamicRange | 2 | 8 | 0.64 | 178.0 |
Ethnicity | 1 | 1 | 120.67 | 149.0 |
Ethnicity | 2 | 1 | 114.94 | 149.0 |
Ethnicity | 2 | 8 | 134.65 | 149.0 |
NIRLivenessEstimator | 1 | 1 | 84.74 | 44.0 |
NIRLivenessEstimator | 2 | 1 | 48.97 | 44.0 |
NIRLivenessEstimator | 2 | 8 | 59.09 | 144.0 |
LivenessRGBMEstimator | 1 | 1 | 235.84 | 142.0 |
LivenessRGBMEstimator | 2 | 1 | 121.32 | 141.0 |
LivenessRGBMEstimator | 2 | 8 | 136.67 | 407.0 |
YUV12toRGB | 1 | 1 | 2.52 | 28.0 |
YUV12toRGB | 2 | 1 | 2.53 | 28.0 |
YUV12toRGB | 2 | 8 | 2.57 | 28.0 |
YUV21toRGB | 1 | 1 | 2.51 | 29.0 |
YUV21toRGB | 2 | 1 | 2.58 | 29.0 |
YUV21toRGB | 2 | 8 | 2.6 | 29.0 |
FaceOcclusion | 1 | 1 | 58.56 | 133.0 |
FaceOcclusion | 2 | 1 | 50.53 | 133.0 |
FaceOcclusion | 2 | 8 | 67.4 | 133.0 |
Rockchip (weak) environment. Detector performance#
The table below shows the performance of Detector on the Rockchip (weak) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
Detector (minFaceSize=20) | 1 | 1 | 13280.8 | 604.0 |
Detector (minFaceSize=20) | 4 | 1 | 6632.62 | 625.0 |
Detector (minFaceSize=20) | 4 | 8 | 6952.72 | 4536.0 |
Detector (minFaceSize=50) | 1 | 1 | 1886.54 | 136.0 |
Detector (minFaceSize=50) | 4 | 1 | 963.36 | 157.0 |
Detector (minFaceSize=50) | 4 | 8 | 998.26 | 829.0 |
Detector (minFaceSize=90) | 1 | 1 | 598.28 | 75.0 |
Detector (minFaceSize=90) | 4 | 1 | 295.25 | 89.0 |
Detector (minFaceSize=90) | 4 | 8 | 321.53 | 344.0 |
Redetect | 1 | 1 | 12.1 | 130.0 |
Redetect | 4 | 1 | 8.21 | 134.0 |
Redetect | 4 | 8 | 7.18 | 787.0 |
Landmarks5Detector | 1 | 1 | 4.37 | 140.0 |
Landmarks5Detector | 4 | 1 | 2.66 | 141.0 |
Landmarks5Detector | 4 | 8 | 2.27 | 141.0 |
Landmarks68Detector | 1 | 1 | 36.15 | 140.0 |
Landmarks68Detector | 4 | 1 | 22.36 | 141.0 |
Landmarks68Detector | 4 | 8 | 19.07 | 141.0 |
Rockchip (weak) environment. Extractor performance#
The table below shows the performance of Extractor on the Rockchip (weak) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
Extractor | 1 | 1 | 8613.25 | 389.0 |
Extractor | 4 | 1 | 4397.11 | 387.0 |
Rockchip (weak) environment. HeadDetector performance#
The table below shows the performance of HeadDetector on the Rockchip (weak) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HeadDetector (minHeadSize=20) | 1 | 1 | 13286.9 | 599.0 |
HeadDetector (minHeadSize=20) | 4 | 1 | 6614.01 | 620.0 |
HeadDetector (minHeadSize=50) | 1 | 1 | 1863.81 | 132.0 |
HeadDetector (minHeadSize=50) | 4 | 1 | 922.59 | 153.0 |
HeadDetector (minHeadSize=50) | 4 | 8 | 992.9 | 830.0 |
HeadDetector (minHeadSize=90) | 1 | 1 | 566.68 | 71.0 |
HeadDetector (minHeadSize=90) | 4 | 1 | 295.05 | 85.0 |
HeadDetector (minHeadSize=90) | 4 | 8 | 322.76 | 345.0 |
Rockchip (weak) environment. HumanDetector performance#
The table below shows the performance of HumanDetector on the Rockchip (weak) environment.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HumanDetector (resize to 320) | 1 | 1 | 305.31 | 56.0 |
HumanDetector (resize to 320) | 4 | 1 | 150.85 | 66.0 |
HumanDetector (resize to 320) | 4 | 8 | 160.52 | 189.0 |
HumanDetector (resize to 640) | 1 | 1 | 1250.47 | 89.0 |
HumanDetector (resize to 640) | 4 | 1 | 630.89 | 103.0 |
HumanDetector (resize to 640) | 4 | 8 | 650.26 | 469.0 |
HumanRedetect | 1 | 1 | 57.47 | 88.0 |
HumanRedetect | 4 | 1 | 31.58 | 98.0 |
HumanRedetect | 4 | 8 | 28.3 | 432.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=20) | 1 | 1 | 14889.2 | 788.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=20) | 4 | 1 | 7502.5 | 812.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=20) | 4 | 8 | 7977.97 | 6050.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=50) | 1 | 1 | 2058.62 | 158.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=50) | 4 | 1 | 1091.52 | 173.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=50) | 4 | 8 | 1142.59 | 1076.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=90) | 1 | 1 | 661.33 | 76.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=90) | 4 | 1 | 340.6 | 91.0 |
HumanFaceDetectorBoxesAndAssociations (minFaceSize=90) | 4 | 8 | 363.05 | 430.0 |
HumanFaceDetectorBoxes (minFaceSize=20) | 1 | 1 | 14223.4 | 732.0 |
HumanFaceDetectorBoxes (minFaceSize=20) | 4 | 1 | 7091.03 | 760.0 |
HumanFaceDetectorBoxes (minFaceSize=20) | 4 | 8 | 7600.78 | 5854.0 |
HumanFaceDetectorBoxes (minFaceSize=50) | 1 | 1 | 2010.47 | 150.0 |
HumanFaceDetectorBoxes (minFaceSize=50) | 4 | 1 | 1043.17 | 165.0 |
HumanFaceDetectorBoxes (minFaceSize=50) | 4 | 8 | 1086.0 | 1016.0 |
HumanFaceDetectorBoxes (minFaceSize=90) | 1 | 1 | 657.1 | 74.0 |
HumanFaceDetectorBoxes (minFaceSize=90) | 4 | 1 | 315.94 | 94.0 |
HumanFaceDetectorBoxes (minFaceSize=90) | 4 | 8 | 346.75 | 396.0 |
HumanWarper | 1 | 1 | 2.56 | 48.0 |
HumanWarper | 4 | 1 | 2.73 | 49.0 |
HumanWarper | 4 | 8 | 1.8 | 90.0 |
HumanWarper | 1 | 1 | 2.93 | 83.0 |
HumanWarper | 4 | 1 | 2.77 | 84.0 |
HumanWarper | 4 | 8 | 1.82 | 125.0 |
Rockchip (weak) Estimations performance without batch interface#
The table below shows the performance of Estimations on the CPU for estimators that do not have a batch interface.
All these measurements are performed with minFaceSize=50
.
Type | CPU threads | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|
LivenessFPR | 4 | 353.95 | 153.0 |
LivenessFPR | 1 | 1199.35 | 150.0 |
PPE | 4 | 88.89 | 100.0 |
PPE | 1 | 277.88 | 100.0 |
Overlap | 4 | 32.46 | 139.0 |
Overlap | 1 | 110.82 | 140.0 |
Rockchip (weak) environment. Estimations performance with batch interface#
The table below shows the performance of Estimations on the Rockchip (weak) environment for estimators that have a batch interface.
Type | CPU threads | Batch Size | Percentile 95 (ms) | RAM Memory (Mb) |
---|---|---|---|---|
HeadPose | 1 | 1 | 3.5 | 139.0 |
HeadPose | 4 | 1 | 2.06 | 138.0 |
HeadPose | 4 | 8 | 1.76 | 180.0 |
Warper | 1 | 1 | 20.44 | 130.0 |
Warper | 4 | 1 | 20.54 | 130.0 |
Warper | 4 | 8 | 7.96 | 134.0 |
Eyes | 1 | 1 | 29.38 | 132.0 |
Eyes | 4 | 1 | 9.26 | 134.0 |
Eyes | 4 | 8 | 9.3 | 138.0 |
Eyes | 1 | 1 | 24.3 | 132.0 |
Eyes | 4 | 1 | 9.26 | 133.0 |
Eyes | 4 | 8 | 9.51 | 138.0 |
Eyes | 1 | 1 | 10.86 | 48.0 |
Eyes | 4 | 1 | 5.7 | 48.0 |
Eyes | 4 | 8 | 6.53 | 50.0 |
Eyes | 1 | 1 | 10.34 | 48.0 |
Eyes | 4 | 1 | 5.81 | 48.0 |
Eyes | 4 | 8 | 6.05 | 50.0 |
InfraRed | 1 | 1 | 58.77 | 51.0 |
InfraRed | 4 | 1 | 27.7 | 50.0 |
InfraRed | 4 | 8 | 18.89 | 71.0 |
AGS | 1 | 1 | 3.79 | 138.0 |
AGS | 4 | 1 | 2.11 | 138.0 |
AGS | 4 | 8 | 1.83 | 180.0 |
BestShotQuality | 1 | 1 | 9.18 | 139.0 |
BestShotQuality | 4 | 1 | 3.85 | 140.0 |
BestShotQuality | 4 | 8 | 2.71 | 181.0 |
MedicalMask | 1 | 1 | 188.66 | 157.0 |
MedicalMask | 4 | 1 | 94.59 | 158.0 |
MedicalMask | 4 | 8 | 79.46 | 200.0 |
LivenessOneShotRGBEstimator | 1 | 1 | 5594.06 | 271.0 |
LivenessOneShotRGBEstimator | 4 | 1 | 2081.39 | 274.0 |
Orientation | 1 | 1 | 191.35 | 37.0 |
Orientation | 4 | 1 | 117.68 | 39.0 |
Orientation | 4 | 8 | 115.67 | 86.0 |
FacialHair | 1 | 1 | 472.92 | 150.0 |
FacialHair | 4 | 1 | 229.6 | 150.0 |
FacialHair | 4 | 8 | 246.54 | 149.0 |
CredibilityCheck | 1 | 1 | 4416.06 | 224.0 |
CredibilityCheck | 4 | 1 | 2282.76 | 224.0 |
BlackWhite | 1 | 1 | 31.55 | 136.0 |
BlackWhite | 4 | 1 | 9.45 | 137.0 |
BlackWhite | 4 | 8 | 8.52 | 139.0 |
NaturalLight | 1 | 1 | 73.6 | 140.0 |
NaturalLight | 4 | 1 | 37.34 | 141.0 |
NaturalLight | 4 | 8 | 30.47 | 141.0 |
PortraitStyle | 1 | 1 | 28.69 | 138.0 |
PortraitStyle | 4 | 1 | 16.51 | 139.0 |
PortraitStyle | 4 | 8 | 13.78 | 180.0 |
FishEye | 1 | 1 | 67.47 | 141.0 |
FishEye | 4 | 1 | 36.53 | 142.0 |
FishEye | 4 | 8 | 36.28 | 142.0 |
EyeBrow | 1 | 1 | 478.45 | 150.0 |
EyeBrow | 4 | 1 | 227.36 | 150.0 |
EyeBrow | 4 | 8 | 241.39 | 150.0 |
HumanAttribute | 1 | 1 | 381.31 | 59.0 |
HumanAttribute | 4 | 1 | 193.87 | 59.0 |
HumanAttribute | 4 | 8 | 183.91 | 75.0 |
RedEye | 1 | 1 | 69.41 | 134.0 |
RedEye | 4 | 1 | 34.43 | 135.0 |
RedEye | 4 | 8 | 32.69 | 136.0 |
HeadWear | 1 | 1 | 137.64 | 150.0 |
HeadWear | 4 | 1 | 70.05 | 149.0 |
HeadWear | 4 | 8 | 54.17 | 150.0 |
Background | 1 | 1 | 28.18 | 138.0 |
Background | 4 | 1 | 16.72 | 138.0 |
Background | 4 | 8 | 13.63 | 180.0 |
Mouth | 1 | 1 | 209.62 | 141.0 |
Mouth | 4 | 1 | 101.53 | 141.0 |
Mouth | 4 | 8 | 101.41 | 141.0 |
Attributes | 1 | 1 | 2477.58 | 182.0 |
Attributes | 4 | 1 | 1252.85 | 182.0 |
Attributes | 4 | 8 | 1274.17 | 274.0 |
Quality | 1 | 1 | 22.3 | 133.0 |
Quality | 4 | 1 | 12.21 | 132.0 |
Quality | 4 | 8 | 12.29 | 133.0 |
Emotions | 1 | 1 | 518.51 | 149.0 |
Emotions | 4 | 1 | 226.84 | 148.0 |
Emotions | 4 | 8 | 245.28 | 149.0 |
EyesGaze | 1 | 1 | 72.56 | 137.0 |
EyesGaze | 4 | 1 | 19.67 | 138.0 |
EyesGaze | 4 | 8 | 19.29 | 139.0 |
Glasses | 1 | 1 | 26.92 | 133.0 |
Glasses | 4 | 1 | 14.24 | 133.0 |
Glasses | 4 | 8 | 13.09 | 134.0 |
LivenessFlyingFaces | 1 | 1 | 318.71 | 155.0 |
LivenessFlyingFaces | 4 | 1 | 147.77 | 176.0 |
LivenessFlyingFaces | 4 | 8 | 138.81 | 213.0 |
DynamicRange | 1 | 1 | 1.55 | 135.0 |
DynamicRange | 4 | 1 | 1.62 | 136.0 |
DynamicRange | 4 | 8 | 0.81 | 177.0 |
Ethnicity | 1 | 1 | 515.58 | 149.0 |
Ethnicity | 4 | 1 | 252.94 | 148.0 |
Ethnicity | 4 | 8 | 243.93 | 150.0 |
NIRLivenessEstimator | 1 | 1 | 347.37 | 44.0 |
NIRLivenessEstimator | 4 | 1 | 186.14 | 45.0 |
NIRLivenessEstimator | 4 | 8 | 200.63 | 144.0 |
LivenessRGBMEstimator | 1 | 1 | 753.06 | 142.0 |
LivenessRGBMEstimator | 4 | 1 | 233.67 | 141.0 |
LivenessRGBMEstimator | 4 | 8 | 239.51 | 406.0 |
FaceOcclusion | 1 | 1 | 172.69 | 133.0 |
FaceOcclusion | 4 | 1 | 99.17 | 134.0 |
FaceOcclusion | 4 | 8 | 110.91 | 133.0 |
## Runtime performance for embedded environment |
Face detection performance depends on input image parameters such as resolution and bit depth as well as the size of the detected face.
Input data characteristics:
- Image resolution: 640x480px;
- Image format: 24 BPP RGB;
The results for minimum batch size and optimal batch size are shown in the tables below. All the intermediate and non-optimal values are omitted.
Face detections are performed using FaceDetV3 NN.
Descriptor size#
Table below shows size of serialized face descriptors to estimate memory requirements.
"Descriptor size"
Face descriptor version | Data size (bytes) | Metadata size (bytes) | Total size |
---|---|---|---|
CNN 56 | 512 | 8 | 520 |
CNN 57 | 512 | 8 | 520 |
CNN 58 | 512 | 8 | 520 |
CNN 59 | 512 | 8 | 520 |
CNN 60 | 512 | 8 | 520 |
CNN 62 | 512 | 8 | 520 |
CNN 65 | 512 | 8 | 520 |
Table below shows size of serialized human descriptors to estimate memory requirements. Human descriptors are used only for reidentification tasks.
"Human descriptor size (used only for reidentification tasks)"
Human descriptor version | Data size (bytes) | Metadata size (bytes) | Total size |
---|---|---|---|
CNN 102 (deprecated) | 2048 | 8 | 2056 |
CNN 103 (deprecated) | 2048 | 8 | 2056 |
CNN 104 (deprecated) | 2048 | 8 | 2056 |
CNN 105 (deprecated) | 512 | 8 | 520 |
CNN 106 (deprecated) | 512 | 8 | 520 |
CNN 107 (deprecated) | 512 | 8 | 520 |
CNN 108 | 512 | 8 | 520 |
CNN 109 (deprecated) | 512 | 8 | 520 |
CNN 110 (deprecated) | 512 | 8 | 520 |
CNN 112 | 512 | 8 | 520 |
CNN 113 | 512 | 8 | 520 |
Metadata includes signature and version information that may be omitted during serialization if the NoSignature flag is specified.
When estimating individual descriptor size in memory or serialization storage requirements with default options, consider using values from the "Total size" column.
When estimating memory requirements for descriptor batches, use values from the "Data size" column instead, since a descriptor batch does not duplicate metadata per descriptor and thus is more memory-efficient.
These numbers are for approximate computation only, since they do not include overhead like memory alignment for accelerated SIMD processing and the like.