Profile Job Results
Jobs
jp2ernn4p
Results Ready
Name
mediapipe_face_face_landmark_detector
Target Device
- QCS8550 (Proxy)
- Android 12
- Qualcomm® Dragonwing™ QCS8550
Creator
ai-hub-support@qti.qualcomm.com
Input Specs
image: uint8[1, 192, 192, 3]Completion Time
8/8/2026, 4:13:17 PM
Versions
- Lite RT: 1.4.5
- QAIRT: v2.45.0.260326154327
- QNN TfLite Delegate: v2.45.0.260326154327
- Android: 13 (TP1A.220624.014)
- AI Hub: aihub-2026.07.31.1
Estimated Inference Time
155 μs
Estimated Peak Memory Usage
0 ‑ 64 MB
Compute Units
NPU
106
| Stage | Time | Memory |
|---|---|---|
First App Load | 449 ms | 31‑32 MB |
Subsequent App Load | 193 ms | 2‑66 MB |
Inference | 155 μs | 0‑64 MB |
| Lite RT | Value |
|---|---|
| number_of_threads | 4 |
| QNN Delegate | Value |
|---|---|
| backend_type | kHtpBackend |
| log_level | kLogLevelWarn |
| htp_options.performance_mode | kHtpBurst |
| htp_options.precision | kHtpFp16 |
| htp_options.optimization_strategy | kHtpOptimizeForInferenceO3 |
| htp_options.useConvHmx | true |
Sign up to run this model on a hosted Qualcomm® device!
Run on device