Appendix A. Specifications#
Runtime performance for mobile environment#
Face detection performance depends on input image parameters such as resolution and bit depth as well as the size of the detected face. The Aurora platform uses mobilenet by default.
Input data characteristics:
- Image resolution: 640x480px;
- Image format: 24 BPP RGB;
Aurora#
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: armv7l Byte Order: Little Endian CPU(s): 4 On-line CPU(s) list: 0-3 Thread(s) per core: 1 Core(s) per socket: 4 Socket(s): 1 Vendor ID: ARM Model: 5 Model name: Cortex-A7 Stepping: r0p5 CPU max MHz: 1267.2000 CPU min MHz: 200.0000 BogoMIPS: 38.40 Flags: swp half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 evtstrm
The number of threads you can find in tables below.
Aurora environment. Matcher performance#
The table below shows the performance of Matcher on the Aurora environment.
| Type | Model | CPU threads | Batch Size | Average (matches/sec) |
|---|---|---|---|---|
| Matcher | 59 | 1 | 1000 | 16597.5 K |
Aurora environment. Extractor performance#
The table below shows the performance of Extractor on the Aurora environment.
| Measurement | Model | Threads | Batch Size | Average (ms) |
|---|---|---|---|---|
| Extractor | 59 | 1 | 1 | 11817.6 |
| Extractor | 59 | auto | 1 | 1462.37 |
| Extractor | 59 | auto | 4 | 1518.88 |
| Extractor | 59 | auto | 8 | 2132.2 |
Aurora environment. Detector performance#
The table below shows the performance of Detector on the Aurora environment.
| Measurement | Threads | Average (ms) |
|---|---|---|
| Detector FaceDetV2 | 1 | 1191.31 / 1034.12 / 5707.42 |
| (Easy/complex/6 faces) | auto | 210.81 / 230.13 / 834.58 |
Aurora environment. Estimations performance with batch interface#
The table below shows the performance of Estimations on the Aurora environment for estimators that have a batch interface.
| Measurement | Threads | BatchSize | Average (ms) |
|---|---|---|---|
| HeadPose | 1 | 1 | 72.46 |
| HeadPose | auto | 1 | 12.16 |
| HeadPose | auto | 8 | 9.97 |
| Eyes (RGB, useStatusPlan=0) | 1 | 1 | 411.33 |
| Eyes (RGB, useStatusPlan=0) | auto | 1 | 67.54 |
| Eyes (RGB, useStatusPlan=0) | auto | 8 | 59.57 |
| Eyes (RGB, useStatusPlan=1) | 1 | 1 | 427.3 |
| Eyes (RGB, useStatusPlan=1) | auto | 1 | 72.67 |
| Eyes (RGB, useStatusPlan=1) | auto | 8 | 57.98 |
| AGS | 1 | 1 | 83.23 |
| AGS | auto | 1 | 173.76 |
| AGS | auto | 8 | 78.04 |
| BestShotQuality | 1 | 1 | 73.43 |
| BestShotQuality | auto | 1 | 12.18 |
| BestShotQuality | auto | 8 | 9.72 |
| MedicalMaskBatch | 1 | 1 | 2948.68 |
| MedicalMaskBatch | auto | 1 | 406.63 |
| MedicalMaskBatch | auto | 8 | 372.05 |
| LivenessOneShotRGBEstimatorBatch | 1 | 1 | 40971.3 |
| LivenessOneShotRGBEstimatorBatch | auto | 1 | 25502.6 |
| LivenessOneShotRGBEstimatorBatch | auto | 8 | 24494.2 |
| Glasses | 1 | 1 | 443.88 |
| Glasses | auto | 1 | 1298.73 |
| Glasses | auto | 8 | 621.69 |
| Mouth | 1 | 1 | 4751.0 |
| Mouth | auto | 1 | 3200.3 |
| Mouth | auto | 8 | 2410.8 |
| FaceOcclusion | 1 | 1 | 3580.21 |
| FaceOcclusion | auto | 1 | 3824.73 |
| FaceOcclusion | auto | 4 | 3292.99 |
| FaceOcclusion | auto | 8 | 2808.52 |
Aurora environment. Estimations performance without batch interface#
The table below shows the performance of Estimations on the Aurora environment for estimators that do not have a batch interface.
| Measurement | Threads | Average (ms) |
|---|---|---|
| Warper | 1 | 26.2 |
| Warper | auto | 33.15 |
| Quality | 1 | 632.68 |
| Quality | auto | 138.09 |
Descriptor size#
The table below shows size of serialized descriptors to estimate memory requirements.
"Descriptor size"
| Descriptor version | Data size (bytes) | Metadata size (bytes) | Total size |
|---|---|---|---|
| CNN 62 | 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.
Feature matrix#
Mobile versions come in two editions: the frontend edition (or FE for short) and the complete edition.
The table below shows FaceEngine features supported by different editions of mobile platform.
"Feature matrix"
| Facility | Module | Complete | Frontend |
|---|---|---|---|
| Core | Yes | Yes | |
| Face detection & alignment | Face detector | Yes | Yes |
| Parameter estimation | BestShotQuality estimation | Yes | Yes |
| Color estimation | Yes | Yes | |
| Eye estimation | Yes | Yes | |
| Head pose estimation | Yes | Yes | |
| AGS estimation | Yes | Yes | |
| LivenessOneShotRGB estimation | Yes | Yes | |
| Medical Mask estimation | Yes | Yes | |
| Quality estimation | Yes | Yes | |
| Mouth estimation | Yes | Yes | |
| Glasses estimation | Yes | Yes | |
| Face descriptors | Descriptor extraction | Yes | No |
| Descriptor matching | Yes | No | |
| Descriptor batching | Yes | No | |
| Descriptor search acceleration | Yes | No |
See file "doc/FeatureMapMobile.htm" for more details.