| Supported Targets | ESP32-S3 | ESP32-P4 |
|---|
Human Face Detect now support one model msr+mnp. It's a two stage model.
First stage model msr predicts some candidates, then every candidate go through the next stage model mnp.
input_shape : h * w * c
msr : 120 * 160 * 3
mnp : 48 * 48 * 3
| preprocess(us) | model(us) | postprocess(us) | |
|---|---|---|---|
| msr_s8_v1_s3 | 10691 | 30922 | 679 |
| msr_s8_v1_p4 | 5509 | 13308 | 352 |
| mnp_s8_v1_s3 | 5101 | 5263 | 105 |
| mnp_s8_v1_p4 | 5021 | 2465 | 41 |